All around the Wrekin

d^mmvobovion
Sodetip's . . .
publications. .

CoTO "^

Jill arouna tDe Ulrekin.

BY

Professor SIR JOHN RHYS, M.A., D.Litt.,

Principal of Jesus College, Oxford.

From "Y Cfmmrodor", Vol. XXI (Printed November 1908).

LONDON : ISSUED BY THE SOCIETY,
NEW STONE BUILDINGS, 64 CHANCERY LANE.

3tl;aca, £?eni fork

BOUGHT WITH THE INCOME OF THE

FISKE ENDOWMENT FUND

THE BEQUEST OF

WILLARD FISKE

LIBRARIAN OF THE UNIVERSITY 1868-1883
1905

The date shows when this volume was taken.

To renew this book pony the call No. and give to
the librarian.

HOME USE RULES

' All Books subject to recall

All borrowers mtist regis-

ter in the library to borrow

books for home use.-

" ■ All books must be re-

I turned at end of college

year for inspection and

'^ ^ repairs.

Limited books must be

• returned within the four

week limit and not renewed.
Students must return all

.^ books before leaving town.

Officers should arrange for

, the return of books wanted

during their absence from

• • town.

Volumes of periodicals

and of pamphlets are held

.in the library as much as
powible. For special pur-

, ^ \ poses they are given out for

a limited time.

Borrowers should not use

their library privileges for

; the benefit of other-persons.

Books of special value

• and gift books, when the

giver wishes it, are ' not

allowed to circulate.
Readers are asked to re-
port all cases of books
""* marked or mutilated.

Do not de^ce books by marks and writiofg.

Cornell University Library
DA 670.S4R47

All around the Wrekin,

3 1924 028 058 059

The original of tliis book is in
tine Cornell University Library.

There are no known copyright restrictions in
the United States on the use of the text.

http://www.archive.org/details/cu31924028058059

Lettsotne^ Llangollen.

Pillar of Eliseg, shewing the Modern Inscription.

Cjmmro&nr*

Vol. XXI. "Caeed boeth yb encilion." 1908.

Bv Pbofessoe sir JOHN RHYS, M.A., D.Litt.

Some two miles and a half to the south-west of Wellington
is the Wrekin, a long and isolated hill which rises some
nine hundred feet above the level of the country round,
except on the north-east, wh§re there is another and a
more irregular hill, called Ercal. They are separated by a,
deep little glen, along which a very pretty brook winds its
way ; the line of the hills is, roughly speaking, north-east
and south-west. The ridge of the Wrekin forms a sort of
long street, except that there are no houses to obstruct the
view on either hand, but here and there plenty of trees.
The whole hill is an ancient stronghold, forming a double
camp two thousand feet long ; the fortifications are now
somewhat effaced in parts, but enough remains to show
that they consisted of a double vallum and fosse, with out-
works. I take these details from the proof-sheets kindly
lent me of the article on "Earthworks", in the first volume
of the Victoria County History of Shropshire; for a full
description of the hill the reader must be referred to the
forthcoming volume, but I have given enough to shew that
the Wrekin is one of the most remarkable fortifications in
the British Isles. That is apart from the fact pointed out

i-L

2 All around the Wrekin.

by Mr, Davies in his Eandhoole to the Wrehin (Shrewsbury,
1895), that this hill is geologically one of our most
primeval landmarks.

I now proceed to quote a passage from Miss Burne's
Folk-lore, Legends and Old Customs, reprinted from her
Memorials of Old Shropshire (Bemrose & Sons, London),
as follows : —

"Wrekin Wakes, held on the first Sunday in May,
were distinguished by an ever-recurring contest between
the colliers and the agricultural population for the posses-
sion of the hill. This is said to have gone on all day,
reinforcements being called up when either side was worsted.
The rites still practised by visitors to the Wrekin doubtless
formed part of the ceremonial of the ancient wake. On the
bare rock at the summit is a natural hollow, known as the
Kaven's Bowl or the Cuckoo's Cup, which is always full of
water, supposed to be placed there as it were miraculously,
for the use of the birds. Every visitor should taste this
water, and, if a young girl ascending the hiU for the first
time, should then scramble down the steep face of the
cliff and squeeze through a natural cleft in the rock
called the Needle's Bye, and believed to have been
formed when the rocks were rent at the Crucifixion ; should
she look back during the task, she will never be married.
Her lover should await her at the further side of the
gap, where he may claim a kiss, or, in default of one, the
forfeit of some article of clothing — a coloured article, such
as a glove, a kerchief, or a ribbon, carefully explained the
lady on whose authority the last detail is given."

Having read this about the Wrekin Wakes some years
ago, I had long wished to make closer acquaintance with
the old camp, and on the 13th of September 1907, in the
interval of two of the many meetings which Welshmen
have to attend at Shrewsbury, I escaped to Wellington,
and had a most agreeable walk to the summit of the
Wrekin, though the latter portion of it was a pretty stiff
climb. One can, however, break the climb at a con-
veniently situated refreshment place on the shoulder of

All around the Wrekin. 3

the hill, before you come in sight of the camp. The
weather was dry, and I was disappointed to find the
Eaven's Bowl empty, but a rock hollow, not far off, held
water still, which my companion's dog found most wel-
come. Perhaps that should have been the Eaven's, and
the other the Cuckoo's, separate provision being made for
the two birds. The most probable view, however, is that
the Cuckoo is to be discarded altogether as a mere intruder
there as elsewhere. Glimpses of many counties may be
caught from the top of the Wrekin, but I am more inter-
ested in a spot only some few miles away, namely, the site
of the Eoman fortress of Viroconium, in English, Wroxe-
ter, on the Severn. For till I visited the "Wrekin I could
never understand why the Eomans built a fortress at
Wroxeter ; but the moment I saw what the Wrekin camp
is like I saw also that Wroxeter was meant to keep it in
check, that is, until it could be made untenable by the
conquest of all the surrounding country. The Wrekin
would not be the sort of nest which the Eomans would
care to occupy any more than the Celts would have elected
to fortify the site of Wroxeter on the level ground. In
Eoman times the inhabitants of the district would seem
to have been the Brythonic tribe of the Cornavii.

I.

If you search the volumes of the Archoeologia Gambren-
sis for the years 1863 (pp. 134-56, 249-54, 334) and 1864
(pp. 62-74, 156-76, 260-62) you will find the record of a
lively controversy between three men of eminence in the
field of history and archaeology, to wit, Edwin Guest,
Thomas Wright, and Thomas Stephens: they have all
passed away. The subjects of the discussion were Viro-
conium, or Uriconium as they called it, the Wrekin, and

the Elegy to Cyndylan in the Red Book of Hergest, a poem

b2

4 All around the Wrekin.

which was subsequently published at length in Skene's
jFowr Ancient Books of Wales, vol. ii, pp. 279-91. The
elegy consists of over a hundred stanzas, and it has been
usually ascribed to Llywarch H^n. Stanza 80 mentions a
place called Dinlle TJreeonn, which Stephens understood
to mean the site of Viroconium, the lie 'place' of its din
'fortress', for of course he regarded the fortress itself as
a thing of the past. Guest and Wright took it to mean
the camp on the Wrekin, and I have no doubt that they
were right. Guest and Stephens agreed in their analysis
of the word Dinlle : they regarded it as a compound,
meaning, literally, a 'fortress place', which Guest inter-
preted as the place of an actual stronghold, that on the
Wrekin, while for Stephens it was the place where a
fortress had been at some time or other previously. It
happens that they were both wrong : not only is their
compound improbable in itself, but we have another
Dinlle, the history of the name of which is clear and easy
to understand. I mean the great mound known as Diuas
Dinlle, on the Arvon coast to the west of the western
mouth of the Menai Straits.

Now the Mabinogi of Math ab Mathonwy informs
us that NantUe, in the same county, took its name
from Llew Llawgyffes, whose older name was Lieu;'
but the Southwallian scribe of the Bed Booh was not
familiar with that name or with the name of Dinlle ;
so when he found Nantilev and Dinftev in his original, he
made them into Nant y tte6 and Dinas Dinttef,' though the
pronunciation meant was Nanttteu and Diniteu, or rather,
perhaps, Nant Lieu and Din Lieu. In fact, it was the
compression of the two words into one, with the accent on
the first, that brought about the shortening of the final

1 Ehys, Hibbert Lectures, pp. 398-400.

2 Rhys & Evans, MMnogion, pp. 71, 78 ; see also ed. note, p. 312.

All around the Wrekin. 5

syllable so as to make the present forms, NantUe and
Dinlle. This gluing together of two words under one
accent is a favourite way of treating place-names in North
Wales : take for example Gastellmarch and Aberffraw. The
surmise as to the old pronunciation of the names in ques-
tion is established by the rhymes in one of the Tomb
Englyns given in the Myvyrian Archaiology, i, 78'', which,
put into a somewhat normalized spelling, runs thus : —

Y bed yngorthir Nantlleu The grave in the upland of NantUe —

Ny 6yr neb y gynnedfeu Nobody knows its properties :

Mabon fab Modron gleu. It is Mabon's, son of swift Modron.

The relation between LUw and Lieu is obscure : possi-
bly Llew was arrived at as the result of a popular tendency
to change Lieu into a more familiar word, and llew, 'a lion',
may have been regarded as quite satisfactory, though the
story of Lieu never gives him the shape of a lion, but, for
a while, that of an eagle. The old form of the name Lieu
should be Llou, and we seem to meet with it in the
Nennian Genealogies, contained in the British Museum
MS., Harleian 3859; see the Gymmrodor, vol. ix, 176,
where we have Louhen map Gvdd gen, that is Llou hen 'LI
the ancient', son of Gruidgen. The latter name was pro-
bably the full compound name of Grwydion, the father of
Lieu, GwydAon itself being the hypocoristic and secondary
formation from the compound ; the latter seems to occur
as Gwydyen in an obscure passage in the Booh of Aneirin,
where we have eryr Gwydyen,^ which, as meaning Gwydion's
Eagle, would exactly describe Lieu his son. The name is

1 Verse xl, Skene, ii, 75, Stephens's Oododin, p. 242. Since the
foregoing was written Professor Anwyl has pointed out another
instance of Gwydyen in the Myvyrian Arch., i, 230*' where one of the
names with which it rhymes is the singular one of Fobyen ; there
is, he tells me, a Caer Bobien between Aberystwyth and Machynlleth.
With Gwydion the Booh of Taliessin (Skene, ii, 158) associates a cer-
tain Gwytheint ; the name occurs as Gwideint in the Life of St.

6 All around the Wrekin.

further reduced to Gwyden, which occurs in the Booh of
Taliessin (Skene, ii, 190, 193) . Further, the name Lieu has
been usually identified by me with the Irish hero, whose
name was Lug Lamfada, 'Lug of the long hand'. Li
Medieval Irish, to which Lug belongs, the genitive was
Loga ; and the Welsh Lou, to which Lieu has been traced,
is the etymological counterpart of Lug, Loga.

We have other instances of vowel-flanked g yielding
Welsh u, not w. The Latin word pugillares, meaning writ-
ing tablets, was borrowed into Welsh, where it appears as
•peullaSr, used in one of the Taliessin poems (Skene, ii, 141)
in the sense of 'books'. There is a still older form, with
ou, namely poulloraur, as a gloss on pugillarem paginam ;
see the Gapella Glosses, edited by Stokes, in Kuhn's
Beitrcege, vii, 393. The next instance I wish to mention is a
native one, meudwy, 'a hermit': the word is to be analysed
into mevr-dwy, meaning 'servus Dei', from dwy for dwyw,
'god', and meu, which has corresponding to it in
Medieval Irish, mug, genitive moga, 'a slave, a thrall'.
The relation between Lieu and Irish Lug, Loga, is exactly
the same as that between meu (in meudwy') and Irish mug,
moga. This is not proof direct of the identity of the
former words, but if you calculate you will find that the
chances against the identity being a mistaken one are
overwhelming, and in matters of etymology you can
seldom obtain a higher order of proof.

Having practically identified Lieu with the Irish Lug
we know where we are and how to proceed further. For

Beuno in the Elucidarium Volume of the Anecdota Oxoniemia, p. 124.
It is there given to the donor of Celynnog Fawr, in Arron, to the
Saint ; in the Record of Carnarvon, pp. 257, 258, it has been printed
Qtoithenit, which is probably less correct.

1 It would be interesting to know whether the pronunciation
tmudioy, that is moydiBy, is still to be heard in Dyfed or Morgannwg
in case of the word forming a part of some obscure place-name.

All around the Wrekin. 7

the latter name occurred as that of Lugus in Gaulish;'
he seems, in fact, to have been one of the most popular
gods of the Continental Celts. Holder, in his AltcelUsch&r
8pr(ichschats, counts no fewer than fourteen towns on the
Continent called after Lugus, from Lyons to Leyden, and
probably dedicated to him as their special divinity. His
citations shew that the oldest form of the city name was
Lugudunon, but as Gaulish seems to have had a tendency,
like that of Welsh, to lay the stress on the penult, it
became Lugdunon, written in Latin Lugdunum. Compare
Holder's Bothmaros from Roto-mdros, and Mogitmdros from
Mogitvr-maros, with vnogitu == Welsh vnoed in Gweithfoed.
Lugudunon is a compound meaning 'the Lieu fortress', 'the
Lug town' ; for duno-n is represented in Welsh by din, of
much the same meaning as its Welsh derivative dinas, *a
fortress, a town or city' ; Irish had the related form dun,
genitive diine, of the same meaning and use, as in Dun-
garvan, Dunlavin and the like, in Anglo-Irish topography.
You will have anticipated my next proposition, that
Din-Lieu is nothing else than the compound Lugu-dunon
resolved into a quasi-compound or syntactical arrange-
ment, meaning 'the fortress of Lieu or Lug'. This
resolution of the old compounds is characteristic of the
later stages of Brythonic : thus an old compound like
Gwyndy is rare in Wales as compared with the looser
name of Ty gwyn, though they mean equally 'the White
House'. So to the fourteen Luguduna on the Continent,
we have practically two to add in this country, one on
the Wrekin and one near the Menai Straits — I have
quite recently heard of traces of a third. The compound
equivalent to Lugudunum would be, in modern Welsh,

' For more notes on Lugus one may consult my sectional address
at the third Congress for the History of Religions, recently held at
Oxford : see the Transactions, vol. ii, pp. 218-24.

8 All around the Wrekin.

Lleudm, and I should not be surprised if it were to be
discovered yet, say, in an obscure passage in one of the
Welsh poets.

At the Lugudunum now called Lyons, the festival of
Lug was probably held on the first day of August, the
month called after the emperor Augustus. On that day
also was dedicated there an altar to Rome and Augustus:'
the identity of the day for the two festivals was doubtless
not the result of accident, and the name of the emperor
was presumably thereby helped not a little to the popu-
larity which it acquired in Gaul. This day fell near a
great harvest day in the Coligny Calendar, namely, the
fourth day of the month of Eivros, approximately August,
called after Rivos, the name probably of the harvest god,
at any rate of the only divinity recognized in the frag-
ments of that document, namely, twice within the month
of Rivros. In Ireland, the feast on the First of August
was called Lugnasad after Lug, Lunasda in Scotland, and
Luanistyn in the Isle of Man ; but in Wales Augustus has
usurped the place of Lieu, so the feast is known as Gwyl
Awst 'the feast of Augustus', for I venture to translate it
so rather than as 'the feast of August'. The English for
it is Lammas, which is explained in the New English
Dictionary as derived from the Old English hldfmoBsse,
that is, literally, 'loaf-mass', for in the early English
Church the first of August, "Festum Sancti Petri ad
Vincula" in the Roman calendar, was "observed as a
harvest festival, at which loaves of bread were consecrated,
made from the first ripe corn". These indications seem
to associate the god Lieu-Lug with the corn harvest.

A fabulous story about the founding of Lyons is given
by the Pseudo-Plutarch, who introduces ravens into it ;
by itself it carries no weight, but coins occur on which
1 Hirsohfeld, Corpus Inacnptionum Latinarum, XIII, i, pp. 227 249.

All around the Wrekin. 9

the genius of Lugudunum is attended by a raven.' Irish
literature represents Lug's son, Cuchulainn, commonly
attended by ravens. This I am prompted to mention in
connection with the Eaven's Bowl, pointed out on the
Wrekin rock, to which Miss Burne calls attention.

The mimic warfare for possession of the Wrekin hill
seems to form a vivid reproduction of more serious strug-
gles in the distant past between the Cornavii and their foes,
whoever they may have been. What may be the explana-
tion of its being fixed on the First of May I do not
know ; but that has always been an important day in the
Celtic calendar. The year began on iVos Galan-gaeaf, 'Night
of the Winter Calends', that is November Eve : second
only in importance to this was Nos Galan-mai, 'Night of
the May Calends', or May Eve. The third great day in the
calendar was the First of August already mentioned ; and
the fourth should be about the First of February, for filling
which Welsh folklore and literature do aot seem to help.
The Irish calendar, however, supplies Saint Bride,^ "chaste
head of Erin's nuns". Her attributes suggest that she
represented an earlier goddess of fire; in that case the
First of February was not badly chosen as the great day of
her cult.

1 See Holder, s.v. Luffudunon, ii, col. 313.

2 Her name in Irish was Brigit, genitive Brigte, but she was
almost singular in being also oaUed Sand Brigit, genitive Sanct
Brigte : so when her cult was imported into Wales her name became
Sanjfreid: it appears so in Evans's Facsimile of the Black Book of
Carmarthen, fo. 42°'- In modern Welsh it is — or should be — Sanffraid,
with the stress onffraid as in Llansanffraid. Sanffreid seems to imply
Saneta Bi-egit where the b had to be softened to v and the name to
become 8ant Vreid : but the contact of the voiceless mute t with v
made the latter also become voiceless. Thus arose Sant Ffreid,
whence Sanffreid, Sanffraid. Pymtheg 'fifteen', often wrongly ex-
plained, is a parallel : pempe-dec- hecame pymp-^eg, whence pymp-theg,

lO All around the Wrekin.

II.

It is now clear, I hope, that Dinlle Ureconn was not
the Welsh name of Viroconium : Dinlle was a distinct
name meaning Luguduno-n, the stronghold of Lug, in
this instance the one on the Wrekin, Urecown,, more cor-
rectly Urecon, being added to prevent its being confounded
with another Dinlle. Urecon it may be pointed out here
was pronounced as a dissyllable Urecon ; in fact, had DMle
not been treated as a feminine we should have had DinUe
Gurecon, with the g developed before u or w according to
the usual Welsh rule, which, however, it is unnecessary to
dwell upon at this point. In Binlle Urecon the latter name
served as that of the district, and we have it in a slightly
different form in a much older manuscript than the Bed
Booh of Hergest.

I allude to a list of the Cities of Britain appended
to the Historia Brittonum, usually associated with the
name of Nennius. Those cities differ in their names
and their numbers in the manuscripts ; but one of them
mentions a Cair Guricon, which appears in another as
Cair Guorcon.' The spelling of this last is due to con-
fusion of the representative of uiro with the prefix which
in Gaulish was uer, as in Vercinqetorvc and VercassiveUau-
nos : in Welsh it became gwor or gwur, modern gm, and in
Irish /er and /or. Now Cair Guricon should be the caer or
fortress of Guricon, just as Cair Geint in the same manu-
script meant the Fortress of Kent. Such Cair Guricon,
that is Cair Guricon, would more correctly be Cair Uricon,
since cair was feminine. This was undoubtedly Viro-
conium, the site of which, near the village of Wroxeter,

1 ]?or both names see Mommsen's Historia Brittonum cum Addita-
mentu Nennii (published in the Chronica Minora Scec IV V VI VID
vol. Ill, i, 211. ■ > > , h

All around the Wrekin. ii

is about three miles from the foot of the Wrekin and
visible from the Dinlle on the top of that hill. Here I
wish to mention that Guricon occurs as a woman's name in
Gurycon Godheu, one of Brychan Brycheiniog's many
daughters enumerated in the Lives of the Gambro-British
Saints, p. 274;' the same lady is called Gwrgon or
Gurgon in the lolo US8., pp. Ill, 120, 140.

Prom an early date in the sixth century vowel flanked
tenues seem to have been mutated, and the pronunciation
of these names was Gwrygon and Gwrgon, although one
went on for centuries writing c, t, p, just as if they had
remained wholly unaffected. This question is to be
touched upon later; here it will suffice to state the
conclusion that what we have taken as a district name
turns out to have been the proper name of a man or
a woman. Naturally the further inference is that the
Cornavii of the locality considered themselves descendants
of a common ancestor or ancestress, whose name was
Guricon, Gurecon, or Gurcon. In that way the personal
name became practically that of the district, which the
local toast in our day describes comprehensively as : "All
friends round the Wrekin". In the days of the Cornavii
they may have called themselves in the plural, Virocones ;
at all events there is no ti-ace of a formation like the
Latin Viroconium. The case is different with the possibly
related name of Ariconivm,, which may be related also to
Arcunia' and Hercynia (SilvaJ. It survives in Welsh as

1 See the "Brychan Documents", carefully edited by the Rev. A.
W. Wade-Evans in the Cymmrodor, xix, 26.

2 Holder's article on this name, and Walde's on quercui (in his
Latin Dictionary), require to be purged of the bogus Welsh words
introduced into them: these latter have been discussed briefly by me
in the Arch. Camb. Journal, 1907, pp. 87-8. As to cychioynnu, meaning
'to rise', add references to the Anecdota Oxoniensia (Jones & Rhys),
pp. 133, 135, 280.

12 All around the Wrekin.

Ergyng, and in English in the district name of Archen&eld.
in Herefordshire. The former is given in the Historia
Brittonum as Ercing, and by GeofErey of Monmouth as
Hergi/n, while in the Liber Landavensis it has a variety
of spellings from Ergm to Ercicg, all pointing back to
some such a form as Ariconio-n, with an i in the second
syllable and a j in the last.

In Dinlle Urecon and Cair Uricon we have a common
element to equate with the Virocon- of the Latin forma-
tion Viroconium; for this seems to be the best attested
spelling. To explain the equation it is to be noticed that
the unaccented syllable vir, that is to say wir, was shortened
into iir, reducing the whole into Urocon-. The next point
to be noticed is that subsequent to the shortening into
Zfro-con-, this had associated with it, and eventually
substituted for it, an alternative Uri-corv-, perhaps also
Zfro-con- ; for the thematic vowel of the first element in
a compound was subject to much fluctuation. Thus our
post-Eoman inscriptions supply such instances as the
following: — Seng-magU and Sene-^magli, Vendesetli and
Vermi-setli, Vendur-magli and Vinne-magli. Compare such
variants in Gaul as Augustodunum and Augustidunum,
Orgetorix and Orgetirix, and others to be found in
Holder's pages. This being so JJriconium may very
possibly have been a real form of the Latin name, but
not so old as Viroconiwn, or even as Uroconium, which
may also have been one of its forms. The manuscripts of
the Antonine Itinerary, and of Ptolemy's Geography,
contain these and some more forms, which cannot be
discussed here.

Other compound names, beginning with viro as their
initial element, will be found given by Holder, but
in all of them viro is the stem of the word for
' man', Welsh gwr, Old Irish fer, modern Irish fear, Latin

All around the Wrekin. 13

vir. Analogy suggests that gwr represents a Gallo-
Brythonic mr6s, plural vvn, which should have given
singular wr, plural gwyr. Gwr may, however, have
obtained its initial g from the plural : in any case the
English Wrekin for Guricon shows no trace of any sound
before the w. So it would seem that the development of
u into gu dates after the coming of the English into the
district, or that, more correctly speaking, the sound was
there but not such as to make itself perceptible to the
English ear. For it is a feature characteristic not only of
Welsh, but of Cornish and Breton likewise, in which our
giur is written gour : the severance of these dialects may
be dated probably some time in the fifth century. The
shortening here in question took place in an unaccented
syllable; I gather that there was primarily another con-
dition, to wit, that the vowel in the next syllable should be
a broad one, 0, u, or a.

In the instances mentioned it was 0, as we have
had only the one element, uiro, to deal with; that
this extended to other words may be inferred from
the fact to be mentioned presently more in detail, that
unaccented ui or ue, followed by a narrow vowel in the
next syllable, is reduced to Welsh u, approximately of the
same sound as German u, not to Welsh w. Once, however,
wi/ro had become gwr, there might be a tendency to extend
the latter beyond its etymological limits, but Welsh Gwriad
for early Uiriatos, where the second * was i, and not
reckoned as a vowel, is not in point : compare the well-
known Irish name Ferad-aeh, later spelling Fearadhach.

In the Idber Landavensis a number of the compounds
involving uindo-s, modern Welsh gwyn 'white, blessed',
begin with gun, such as Gwnda, from Umdo-tamos, Ghmguas
from Vimdo-missos, Gunva from Umdo-magus, and the
Bishop of LlandafE's palace is called St. Teilo's Gwndy

14 All around the Wrekin.

(p. 120), as if it were Uindo-tegos 'White House'. Most
names of the kind are liable in book Welsh to have the
y of gwyn re-inserted. We have an instance which has
resisted this kind of 'correction' in the name of the Car-
diganshire church of Llanwnnws or Gwnnws, probably
from Uindo-gustus, but the s of Gwnnws for st looks like a
touch of Goidelic influence. One may here also quote
from one of the MSS. of the Mistoria Brittonwm, loc. cit.,
p. 193, the name of Gwrtheym's grandfather, Guttolion,
derived from Vitalianus, which occurs on one of the
bilingual monuments at Nevern, in Pembrokeshire.

But this phonetic change is by no means confined to
the vocables just mentioned ; we have it in forms of great
antiquity, representing the Indo-European perfect of one
of our few strong verbs. The Mahinogion, for instance,
have the following forms, gwdom, gwdam 'we know',
gwdawch, gwdoch 'you know', gwdant 'they know';' since
the Middle Ages they have y inserted after the analogy
of the other forms of that verb, such as gwydwn 'I knew',
gwyhyd 'will know', and gwyhod' 'the fact of knowing,
knowledge'.

' I am indebted for a tabular survey of the tenses of the verb in
question, which occur in the Mahinogion, to Prof. J. Moms Jones, one
of whose pupils is preparing to publish on the verbal forms in those
tales. I should add to them gwdost, ' knowest ', which I cannot ex-
plain. Mod. Welsh gvyddost, in Breton gouzoud. The first person
singular was gwnn, now written gion, which looks like a contraction of
the form which has yielded Breton gouzonn, rather than derived from
a verb corresponding to Irish finnaim 'I find, I know'.

2 This implies uidi-bot- or uide-bot- with the thematic vowel
dropped before the d and b were mutated ; so uid-bot- yielded idpot-,
gwybod; but there was apparently a later compound with the con-
sonants mutated and yielding gwydfod ' immediate personal presence'
— yn ei wff^od = gn ei wgd 'within his knowledge or consciousness
as derived from his sense of sight, hearing, and touch'. The etymo-
logical equivalent in Breton seems to be gouzoud 'the fact of knowing'-
and the compounds with the verb 'to be' are on the same level for

All around the Wrekin. 15

The corresponding forms in the kindred languages
make the structure of our Maiinogion verb at once
intelligible : take Sanskrit veda, Greek olSa 'Iknow',
Sanskrit plural yidma, Greek 'iSfiev 'we know'. Here the
root part of the verb appears in its strongest form in the
singular, while in the plural it is in its weakest ; Sanskrit,
moreover, represents the old accentuation, which explains
the Brythonic gwdom, for instance, as standing for some
such a form as uid-o-mos^ which was weakened into vdomSs,
whence, when penultimate accentuation became the rule,
udomo and (g)4dom, gwdom. The treatment was the same
in the second and third persons of the plural ; and so in
Breton, where the corresponding persons are (1) gouzomp,
(2) gouzoc'h, (3) gouzont; in Cornish (1) godhon, (2)
godhough, (3) godhons ; but, according to Jenner's Hand-
hook of the Garnish Language, pp. 147-8, from which I copy,
godh- has been spread almost over the whole of the con-
jugation.

This explains the etymological difference between
the perfect goruc or gorug, and goreu 'did, fecit'. The
former has by its side gorugum 'I did', and gorugost
'thou didst', but when this stem invaded the plural in
such forms as gorugam 'we did', and gorugant 'they did',
it was encroaching on the domain of goreu-, which, in its

instance goufenn 'I should know', probably for ffouz-venn, and so in the
case of afzjnaout = Welsh adnabod 'to be acquainted with', as to which
see my Celtic Insoriptions of France and, Italy, p. 9. The thematic
vowel belonging to the first part of gvjyhod and gwydfod was probably
i or e which we have in the Latin cognate verb vide-o. It emerges as
iin the Medieval Welsh form gwyctyvm 'I knew, je savais', gwydyei
(Skene ii, 69), and gwydyad 'he knew, ilsavaW : compare the Cornish
godhyen, godhya, and see Norris's Ancient Cornish Drama, ii, 263, 267.

1 As to some of the difficulties connected with the plurals of verbs
of the perfect tense, such as the connecting vowel, the unmutated m
and similar questions, see Brugmann's Orundriss, ii, 1205-7, 1212,
1245-9, 1354.

1 6 All around the Wrekin.

turn, should not have appeared in the singular, but only-
help to make up such a form as gorevMm 'we did' for an
early uo-(u)rogom6s, whence uo^ogom, (g)uo-rog6m, guo-
rouom, gormom, or goreuam. Gorewom and goreuant are
not known to occur, for the reason, perhaps, that they have
not been looked for. In the singular, not only was the
root vowel lengthened, but the mute consonant was pro-
vected;^ both are processes which were probably carried
out under the stress accent. Thus, the third person
singular set out from uo-(u)roce, whence uo-roce, guo-mce,
guoruc, gorug. The corresponding Old Cornish was gwruk,
wruh, rule, rug, later gwrig 'did'. The present tense of
this verb in Welsh occurs in the compound cy^weiriaf 'I
put into working order', from the root verg, and is of the
same conjugation as the Old Irish do-airci (for do-vaircij
'efBects, prepares', Anglo-Saxon wyrcan 'to work, to build'.*
A shortening before the stress syllable, parallel to that
of uiro into uro, has taken place in the name TJrien, written
JJrhgen in the Ristoria Brittonum {he. cit. 63), the same name
most likely as that of the Helvetian pagus mentioned by
Caesar (i, 27) as Verbigenus. We have the Irish form possibly
in the proper name Fergen, in case that represents Ferbgen.
Another instance is Welsh uceint, now ugain 'twenty',
which points back to uicention; the Irish wasjfeAe 'twenty',
genitive fichet. We seem to have a third instance in
Welsh ucher 'evening', from uecsero-s = ueqsero-s, for
uesqueros of the same origin as Greek eowepo<s and Latin
vesper 'the evening'. The Old Irish was fescor, rtowfeascar
'evening'. All these cases differ from the previous ones,
in the contraction being not into w, but into the very

1 For instances of such provection see a paper of mine in the
Xevue Celtique, ii, 331-3.

^ See the Ctrammatioa Celtica, pp. 591-3 ; Jenner, pp. 129-31 ■
Stokes's UrkeltiscJier Sprachschatz — s.v. verg 'to work', p. 273.

All around the Wrekin. 17

different vowel m: the probable explanation is that here
the accented syllable had the narrow vowel e, which
exercised an umlauting influence on the foregoing syllable.
None of these, it will be noticed, shows any trace of an
initial g in Welsh.

III.

Before proceeding any further, I wish to say a word
on early Celtic accentuation and desinence. The former
is not infrequently assumed to have been the same in
Brythonic as in Groidelio, but nothing could be more mis-
taken. In both, it is true, the accent, as far back as we
can trace it, was a stress accent, but in Goidelic it was fixed
on the first syllable in nouns and adjectives, while in
Brythonic it had only the range of the three last syllables
as in Greek. The older accentuation of Latin' appears to
have been on the first syllable, as in Goidelic, but in the
historical period it is found confined to the last three
syllables, as in Brythonic, which was probably the case also
with Gaulish. Within the three-syllable limit, Brythonic
— also probably Gaulish — tended to drive the accent to
the penultimate, and by so doing to put an end to both
oxytones and proparoxytones. The former would, in any
case, be probably few, containing among their number the
vwo-s 'man' already mentioned. The latter were common
enough in Gaulish in such names as the following, where
the position of the accent is practically indicated by the
forms taken in French by such place-names as ArgentS-
magus 'Argenton', Glaudio-magus 'Clion', Novio-magus
'Nyon and Noyon', Boto-magus 'Rouen', Gamho-ritvm,

' Did the Umbro-Samnites, the neighbours of the Romans, accent
their words only within the last three syllables ? and, if so, had their
influence anything to' do with the change of accentuation in Latin ?

1 8 All around the Wrekin.

'Chambort', NoviS^itum 'Niort and Fort'.' In Brj-
thonic we have instances in such names as Brigo-maglog,
Bridmail, Bridfael, and the like to be mentioned presently.
Some of the proparoxytones might have penultimates
with longvowels: take, for instance. Oaf w^jgres anABitu-riges,
whence the French place-names Charges and Bowrges. But
such a form as Bitu-riges may have had a tendency to become
Bitu-riges, which seems to be re-echoed in the province name
Berry. ^ Similarly Lugdwno-n, if it was Gaulish, must have
superseded the longer form, which was probably accented
Imgvr-duno-n, and later Lugu-duno-n, before the pretonic
part of the word was curtailed. A good instance of this
occurs in the case of the Gaulish preposition are, in Welsh
ar 'on, upon, at, irapd, irapai', as a prefix in the Gaulish
Aremorica, probably Aremorica, reduced early to Armorica
— the manuscripts of Csesar de Ballo Gallico show no trace
of the pretonic e. The same shortening is attested by the
Gaulish man's name Atpomarus, as compared with the more
usual form Atepomaros, to be mentioned again presently.
Holder, in his AUceltischer 8prachschate, i, 224, has an
Artegia, which is now Arthies in the department of Seine-et-
Oise: this stands for Are-tegia, where tegia represents
tegiha := tegisa, the neuter plural of tegos ' a house or hut ',
Old Irish tech, Welsh ty 'house'. With the Gaulish pre-
position translated into Latin ad we have ad tegia and ad
teia, which appears to have entered the place-name* Adtegia,
now called Athies, in the department of the Somme, and a
common noun attegia 'a hut or tent', not to mention that
tegia survives, for instance, in the Tyrol as thei, tai *an

1 See Meyer-Liibke in the Sitzungsberichte der kais. Akademie der
Wissenschaften in Wien, oxlii, ii, 40, 44 ; see also the separate
names in Holder's AUceltischer Sprachschatz, which is arranged
alphabetically.

2 See, however, Meyer-Liibke, loc. cit, p. 10. "

All around the Wrekin. 19

Alpine hut', with which compare the Welsh iai 'houses',
Med. Welsh lei for tegia from iegesa.^

One or two other instances will help to illustrate the
difference between Irish and Welsh with regard to accen-
tuation. One of the words in point is the Old Irish neuter
dorus 'a door', from some such a stem as duorestii-, in
Welsh drws from duorosf/u-, which must have been accented
duorSstu, otherwise the first syllable could not have been
reduced to the consonants dr: compare Graulish Duro-
casses, yielding in French the place-name Dreux. In Irish
this could not have happened, as the stress accent
would there be on the first syllable. A similar instance
offers itself ,in the name of the Denbighshire church
and town of Llanrwst, that is the llan of Gurgust.
When the second g of that name was dropped, the
pronunciation became monosyllabic Gwrwst or Gurust,
which, when preceded by the feminine llan = landa, became
Llanurust, whence the modern pronunciation of Llan 'rwst.
The original compound was Uiro-gustu-s, which made
Uro-gustu-s, and, subject to the tendency of the accent to
rest on the penultimate, became (G)uro-gustu-s, and later
Gurgust. For Irish the compound was Vira-gustu-s, bub
being accented on the first syllable the resultant form is
the well-known name Fergus.

The next instance to be mentioned is one in which
I cannot vouch for the correct sequence of the phono-
logical modifications involved : Old Irish had a neuter
noun aithesc, which comes from dti-segua-n, which
became dthesgua-n, cUthesc-n, For Brythonic this would

1 See Mayer-Liibke, loa. cit., pp. 12-13, who has been improved on
by Holder in several respects ; but from not knowing that tegia was
etymologically a plural itself, he has suggested ad tegia(») and are
tegia(s), with an s, which the authors of most of the old documents
to which he refers did not think necessary. See also Walde's
Lateinisches eti/moloffisehes Worterbuch, a. v. attegia.

2

20 All around the Wrekin.

be ate-hepo-n, probably ate-hepo-n, whence ade-h&po-n, ode-
hep, ad-hep, dtep, dteh 'answer'. We have possibly traces
of this word in Gaulish : Holder gives two proper names,
Atepomaros and Ateporlx. They are usually explained with
some trouble, with the aid of the Gaulish epos 'a horse';
but we have so much 'horse' in Gaulish nomenclature
that it is a relief to find something else. Should the
conjecture that atepo-n (for ate-hepo-n) enters into those
two names, the compounds must have meant respectively,
'One who is great in his replies' and 'One who answers
like a king'. It is needless to say that those great names
had shortened and hypocoristic forms: one of these
Holder gives as Atepihs, and from Latin contexts Atepa,
Atepatus, Atepiccus, Atepilla and Atepo, genitive Aleponis.
A Gaulish parallel to atepo-n would be ar&po-n^ from
are-hepo-n. I have no proof of its having existed, but in
Irish we have its counterpart in airesc 'a saying', in
Welsh di-areh 'a proverb', now pronounced dihdreb, plural
dia/rhehion.

There is no need to dwell in general terms on the con-
nection between the case endings of a word and the accent
which falls in that direction, as it did in Brythonic.

1. One of the points of principal importance to notice
is the fact that the endings of the nominative case in the
vowel declensions o-s, u-s, t-s, fell away so early that they
have not perceptibly affected our mutation system in Bry-
thonic.

' This reminds me that Holder has are-pos suggested by the re-
versible words : SATOR They will be found in the Berlin C. I. L.,
AREPO xii, 202*, where it is suggested that they
TENET are not earlier than the seventh century.
OPERA Holder mentions two translations which
ROTAS have been proposed of the puzzle ; they are :
o aireipav Uporpov Kparei cfyya rpoxovs, and "Le laboureur Arepo
tient aveo soin les roues".

All arottnd the Wrekin. 21

2. There is no apparent reason why this remark should
be limited to the nominative endings just mentioned : it is
probable that their history was bound up with that of the
other short-vowel endings ; that is, they were all swept
away by the same phonological tendency, and in the same
period. The principal endings in point would be the
vocative singular e of the declension, the a of the
neuter plural in the nominative and accusative of all
declensions, the o-s of the genitive singular of the con-
sonantal declensions, the e-s of the nominative plural,
masculine and feminine of the same declensions, and the
e of the nominative and accusative dual in the same.'

3. On the other hand, the long-vowel endings are
supposed to have lasted longer, so that while the others
were wholly dropped the long vowel was only curtailed,
not completely dropped, for some time later. Thus, while
in the masculine mndo-s became (g)uind, gwynn, gwyn
'white', the feminine uindd only became uenda, whence
later (g)uend, gwenn, gwen. At all events the feminine
ending a aa a remained long enough to leave its mark
permanently on our mutation system. Take a common
instance like the feminine llaw goch 'a red hand', derived
Erom lama cocea, the c between the two vowels being
mutated to g by the influence of those vowels. Other
instances would be the genitive singular of the declen-
sion, which ended in J like the Latin domint, the 5 (or u)
of the dative of that declension, like Latin domind, and
the nominative plural in I like Latin domini. To these
should be added the ending 5 of the nominative, vocative,
and accusative of the dual in the declension, and of the
genitive dual in all the declensions. The vowels in question

' A glance at Stokes's Celtic Declension, especially his tables,
pp. 100-04, or those in Brugmann's Grundriss, ii, 736-59, will make all
this clear.

22 All around the Wrekin.

were probably reduced to i, o ox u before they ceased
altogether to be pronounced, which took place late enough
for them to have affected the mutation system. Why they
did not do so in the case of the plural is explained by the
endings : there was a lack of unanimity to establish a
mutation : the nominative plural of the declension, for
instance, ended in i, while the corresponding feminine had
as and the consonantal declensions es. Not so with the
dual, which, though comparatively little used, has left the
soft mutation to mark its presence in the background even
in Modern Welsh : witness, for instance, the Welsh word-
ing of the Church of England's bans of marriage, where
we have y deudyn hyn 'these two persons' : here the
softened d, in both instances, is due to the ancient dual.
For that number had a vowel termination in all the cases
except the dative, which had a dissyllabic ending : this is
not quite certain. But the others agreed in leading up to
the soft mutation, and a remarkable instance offers itself
in the elegy, already mentioned, to Cyndylan, stanza 28,
where we have the following lines : — *

Stauell gyndylan yspeitha6c [?] heno

g6edy ketwyr uoda6c
Eluan kyndylan kaea6c.

"Cyndylan's chamber, it is desolate to-night :
Gone the two contented warriors,
Blvan and torque-wearing Gyndylan."

1 See Skene's Four Ano. Books of Wales, i, 452, ii, 282, 445. In his
notes Skene writes as follows: — "The first 57 stanzas of this poem
have been carefully translated by Dr. Guest in the Archeeologia Cam-
brensis, ix, p. 142, and the translation has been, with his permission,
adopted. The reader is referred to the notes by Dr. Guest on this
part of the poem. The remaining stanzas have been translated by
Mr. Silvan Evans." In this instance, Skene's process of 'adopting'
Guest's translation involves changing the latter's "contented" into
"contended", and misrepresenting the sense of the original* for
Guest was practically right here, though he was not by any means

All around the Wrekin. 23

The words in question more particularly are ketwyr
uodadc, which seem to point back to an early combination
catvr-idro bodoco which, as regards the case ending of the
dual, might be nominative, accusative, or genitive. The
preposition guedy 'after' should decide, but it is not known
what case it governed. In Old Welsh it is found as guetig
and guotig,^ but the etymology is obscure. If it involves
a nominal element it probably governed the genitive ; of
the three cases, the only other one which the sense would
seem to admit is the accusative, which appears less likely
than the genitive.

We may now examine the alternative forms Qwricon
and Gurcon from the point of view of their etymology, so
as to shew in what sense they are entitled to be regarded
as equivalents. It happens that we have the exact
equivalent of Gurcon or Gv/rgon, in the Irish name
Ferchon, which is nought else than the genitive of a
compound which is in the nominative Ferchu,' to which
corresponds exactly the Old Welsh Gurcu in the Liber
Landavensis, later Gurci, sounded Gwrgi : it is matched by
Gurcon in the same manuscript, which supplies a number
of other similar instances, such as Flcu or Elci, and Elcwn
or Ficon, Guidci and Gvddcon. But though those ending
in con or cun were, etymologically speaking, the genitives
of those ending with cm, ci, they are there treated as
distinct names. This would have been impossible here in

equal to the task he had undertaken. If Silvan Evans had trans-
lated the 67 stanzas we should have had a correct rendering of the
portions then intelligible to a man well trained in literary Welsh.
Skene, however, does not appear to have known enough Welsh to
help him to judge correctly as to their respective merits in the
matter of translating.

1 See the Grammatica Celtica, p. eSS""-

2 See Windisch's Tdin B6 CHailnge, 2,893, 2,914, and The Book of
the Bun Cow, f. 82''-

24 All around the Wrekin.

Old Irish, as Ferchon would at once be associated with
Ferchu, cu, genitive' con, being words familiar to all who
spoke Irish. It was different in a language where, as in
Brythonic, the system of case-endings had gone to pieces.
So we find the same thing happening in other instances :
take, for example, the Latin word for city or state, civitas,
genitive civitatis ; in Welsh the one yielded regularly
eiwed and the other ciwdawd or ciwdod. Here the
language has utilized both; ciwed has now the sense of
'a rabble', and ciwdod that of the people or population of
a city. We have another instance in trined and trindod,
from Latin trinitas, genitive trinitatis 'a trinity'. Here
the language, having seemingly found no special use for
trined, lets it become obsolete. Lastly, we have a native
instance in Ghuyned and Crwyndod (for Gwyndot), from an
early Venedos, genitive Venedotos, which occurs in a Latin
inscription as Venedotis, to wit, at Penmachno in Carnarvon-
shire. Gwynett is the form in ordinary use, while Gwyndod
is left to the poets, and to be the base for Gwyndodes 'a,
Venedotian woman', and Gwyndodeg 'the Venedotian
dialect of Welsh'.

Similarly, the accent has left us a certain number
of compound proper names with two forms each, as
Urbagen or Urbeghen, and Urbgen,^ later Urien; Tutagual
and Tudwal ; Dumnagval and Dyfnwal ; Dinogat or Dinagat
and Dingad. The early nominatives of these last were
Toutovalos, Dubnovalos or Bumndvalos, and Diinocatvs, to
which may be added Brigomaglos, which became later
Bridmail, Bridfael. This accentuation has been proved in
the case of names of similar composition, and the same
number of syllables in Gaulish; see p. 17 above. But, as

1 See the Historia Brittonum, loc. cif., 206-7 ; Nicholson's "Filius
Urbagen" in Meyer & Stern's Zeitschrift fiir celtische Philologie,
iii, 104-11.

All around the Wrekin. 25

in our instances the endings -os and -us were discarded
early, the nominatives became, for example, Toutoval and
DunScat, which provided a stable position for the accent.
That is proved by the later forms being Tutdgual (or
Tuddwal) and Dinogat (or Dvndgat), without any shifting
of the accent. This would apply probably also to the
corresponding Brythonic accusatives, To-utovalon and
Dunocatun ; but when we come to case-endings with a long
vowel, which would remain longer intact, a shifting of the
accent probably took place : thus the genitives Tout6-vali
and Du7i6-catdus or Buno-catos, became probably Touto-vdli
and Duno-cdtos, whence resulted Tout-udli, Bwn-gdtos,
whence Tududl, Dingdt, and later, Tudwal, Dmgad. The
resulting forms in the dative, ablative, locative, and
instrumental would, if they existed, be probably identical.
One of the steps here guessed, namely, that from Touto-valt,
let us say, to Tout-udli, recalls a Gaulish proper name
already mentioned as Atepilos, that is probably Atepilos.
We seem to meet with its genitive variously written AtpiU
and Atpilli, which were accented, probably AtpiU, Aipilli.
See Holder s.v. Atpillos, Atpilos, nominatives for which, be
it observed, he cites no authority.

The foregoing instances belong to the declension
(Toutovalos) and the U declension CDunocatus) ; when we
come to the consonantal declension it is not so clear what
has happened, but the same general rules of accentuation
may be assumed to have applied. The results, however,
differ conspicuously from those in the vowel declen-
sions, for here we may have not two forms but three.
Unfortunately the names to our purpose are only two :
they have both been already partly discussed, G-v/rcu and
Mailcu. The nominatives must have been Uirocu,
Maglocu, accented probably on the cu ; this would lead to
the elision of the immediately preceding the stress

26 Alt around the Wrekin.

syllable, and, with the consonants softened previously, we
should have [Glwrgii (written [G]wrcu.), Gwrgi, Gwrgi (written
Gurci). Similarly with Mailcu, Elcu, and the like. Next
comes the genitive^ which should have been Uirocunos or
Uiroconos, reduced to Uroconos, with optional forms
Ureconos or Urieonos. These fall into the same accentua-
tion as Brigomaglos, ToutSvalos, and the like, yielding
accordingly Vreconos or Urieonos, and, when the short-
vowel case ending went, Urecon or TJricon, whence the
attested forms Ureconn, Gwricon, Gurycon. There remains
Gurcon, which may be explained in one of two ways.
(1) The gwr of Gurcon may be due simply to the analogy
of Gurcu in the nominative, and the formation may have
been meant as a genitive, which in due course superseded
Guricon. (2) It is, on the whole, more probable that it
represents another case, say, the dative. So we set out from
TJrSconl with a final t as in Latin hominl, and assume that
it would take longer time for the i to be dropped than in
the case of a short-vowel termination. So we may set
down JJrgSm as the next stage, whence one arrives at
TJrgSn, Gurgon, Gwrgon (written Gurcon).

One would reason similarly as to Mailcon or Mailcvn,
and we have a trace of the genitive as Meilochon in Brvde
mac Meilochon, the name of more than one Pictish king :
the father of the first of that name has sometimes
been supposed to have been Maelgwn, king of Gwyned.
It is remarkable that B. mac Meilochon comes in Bede's
Ecclesiastical History, iii, 4 : in Irish annals it is more
usually mac Mailcon or mac Maelchon. In Meilochon,
as well as in Maelchon, the ch is an Irish touch, which
must be due to the scribe who first wrote it in this name
being aware of the fact that in Brythonic the original c
was mutated to g, whether written so or not, and that the
corresponding Irish mutation was to ch, which he accord-

All around the Wrekin. 27

ingly used in his spelling of this genitive, Meilochon : that
is to say, he knew that the Brythonic pronunciation was
Mailogon, probably Mailogon ; we have possibly the same
formation in Breton, to wit, in Maelucun, which occurs in
the Cartulary of Landevennec, published by MM. Le Men
and Ernault. Gildas, addressing Maelgwn in the voca-
tive, calls him Maglocune, which suggests that he would
have used Maglocunus as the nominative in Latin. With
this agrees the bilingual inscription lately discovered at
Nevern, in which the Latin genitive is Maglocuni, though
the Goidelic genitive is Maglicunas.^ It is interesting to
find GeofBrey of Monmouth producing a faint echo of the
purely Brythonic declension of the name in his Malgo,
genitive Malgonis, accusative Malgonem.

On looking back at our conclusions, which have been
drawn from the foregoing instances, we seem at first sight
to have a difiiculty in the fact that the longer forms
Dinocat, and Tutdgual, appear to have been nominatives,
and the short ones Bingat, Dingad (as in Llan Dmgad) and
Tutgwal, Tudwal (as in Tnys Tudwal) to have been, let iis
say, genitives, while Guricon or Gurecon, and Meilochon,
that is, Mailogon must be genitive, and the shorter ones,
Gwcu, Gwrgi, and Mailcu, Elcu, nominatives. There is no
real difficulty; it has been shown practically that the
former belong to the vocalic declensions and the latter to
the consonantal ones. The discrepancy between them was
connected with the break up of the older and fuller in-
flection of the noun. In fact, this difEerence of declension
was possibly one of the things which helped to accelerate
that result. The state of things which this indicates
may be appositely compared to what happened in Old
French when the Latin declensional system broke up.
There one finds, for example, the cas regime of the mascu-
1 See the Archceologia Cambrensis, 1907, p. 84.

28 All around the Wrekin.

line singular identical in form with the cas sujet of the
plural, and often enough the cas svjet of the masculine
singular with the cas regime of the plural.' The question
how the declensional system in Brythonic disappeared is
one of great difficulty, owing chiefly to a great scarcity of
data ; but, in fact, the few data available have never been
studied and forced to give up their latent evidence.

The Nevern Ogam, with the genitive Maglicunas, proves
beyond doubt that the second element is the word for
'dog', nominative cu, genitive cunas, dative cwni, which in
Brythonic were probably cimos, cum. In Celtic names
this word had the secondary meaning of guardian,
champion, or protector: so Uiro-cu, Ghircu, Irish Ferchu,
would mean, literally, a 'man guardian' or 'man protector'.
In the other compound, the one with maglo-s, Modern
Welsh mael, and Irish mdl 'a nobleman, a prince, a king',
that vocable is supposed to come from the same root as
Greek /leyaXr), Gothic miJeils 'great', and Scotch michle
'great, much'. In Irish annals the name should appear
as Mdlchu, genitive Mdlchon, but I have no note of meet-
ing with an instance except in the Nevern Ogam. The
name should mean a ' prince guardian' or 'king protector'.
This use of the word for dog or hound in Celtic personal
names is very remarkable, and is borne out by Celtic
history : the Gauls, for instance, used dogs in their wars,
and Strabo tells us that dogs fit for hunting and for war
used to be exported to Gaul from this country. The Irish
word cu is epicene, and in Welsh names it is not restricted
to men : witness Gwrgon and Gurycon as the name of one
of Brychan's daughters already mentioned, to which may
be added from the Book ofLlan Ddv a Leucu (Hiugel's wife),
p. 236, later Lleuci.' So with y Weilgi 'the wolf-dog', as a

1 See Nyrop's Orammaire historique de la Langiie franqaise, ii, 184-9.
' D. ab Gwilym, poem clxvi, has Lleucu, however, to rhyme with

All around the Wrekin. 29

poetic term for the sea, which, though of the same com-
position as the Irish man's name Faelchu, is a feminine.'

IV.

A word must now be said of the English forms of the
name in question, and here I am very pleased to acknow-
ledge my complete indebtedness to the kindness of
Mr. Stevenson, the learned editor of Asser's Life of King
Alfred. According to him WreJein derives directly from
Wreoeen, which he treats as a Mercian modification of
an original Wrekun or Wrikim, the form taken in Old
English by WriJcon, that is the Celtic Uricon. The name
Wrocwardine is, in its first part, of the same origin, and
represents what must have been in Old English Wreocen-
weordign "Wrekin village or Wrekin farm". This became
successively what is found written WroJcewwrdin or (with
Norman ch = h) Wrochewurdi/n, later Wrochwwrdin or
WroewurSin : that is, Wreoeen is first reduced to Wroke,
and then to Wroc, in the "compound. The case of
Wroxeter must have been partly similar. For, setting
out from Wreocen-ceaster, we get a form written Wrocce-
eestre, and French influence makes cestre into sestre, so one
arrives at Wrockesestre, which readily becomes Wroxeter.'

The English form Wrekin, and the others derived from
the same Celtic original, suggest conclusions as to that

I>yddgu, in which the second syllable possibly represents cu 'dear,
beloved'. But in any case one is tempted to ask why Lleucu is not
modified into Zleuci, Lleugu, or Lleugi. The same is the case with
gwenci, a feminine, which is the word in North Cardiganshire for a
weasel.

1 See the Blaoh Book of Carmarthen, f . 38b., and Skene, ii, 40.
In the curious passage about the river fabled to have once separated
Britain and Ireland, y teymassoed should be emended into y
theyrnassoed 'her realms': see the Oxford Mabinogion, p. 35.

* As Mr. Stevenson's monograph is rather too long for a footnote,
it will be found printed at length at the end of this paper.

30 All around the Wrekin.

original which are of interest from the point of view of
Brythonic phonology. Setting out from Uirocoi^, we tnow
that before it was adopted by the English uiro had not
only become uro, but uro and its alternative iiri or ure had
further become monosyllabic, wo, wri. This latter process
of shortening may be dated as near as you like to the
conquest of the Wrekin district by the English, provided
it be treated as dating before that conquest and not after
it. The antecedent change of mro into iiro occurs beyond
Welsh in the Breton language, where the word spelt in
modern Welsh gwr 'a man, vir' is written gow. In other
terms we may probably regard uro for uiro as common
Brythonic, and an accomplished fact before the separation
of Welsh and Breton, say some time in the fifth century.
In the other direction it had not taken place at the time
when the Romans first became acquainted with the
Cornavii of the district. This can hardly have been later
than the presence in this country of the Roman general
Ostorius Scapula, who received command here in the year
50, and proceeded, among other things, to maintain a
boundary extending from the Severn to the basin of the
Trent. It may be guessed to have reached from the site
of Viroconium to that of Pennocrucium. In fact it is
possible that Ostorius it was that selected the former
site and began to fortify it.

The next point of importance to be mentioned is that
when the English borrowed the word which became
Wrekin, the Brythons had not as yet mutated the vowel-
flanked c into g, otherwise the Old English Wreocen would
not have c or k, but g, or else a sound derived from g.
One naturally asks next when did the English first become
familiar with the district and its name : no certain answer
has ever been given that question. It is true that an
entry in the Saaion Chronicle has been supposed by some

All around the Wrekin. 31

to supply it. Under the year 584 we read to the following
effect: — "In this year Ceawlin and Cutha fought against
the Britons at the place which is named Fethanleag, and
Cutha was there slain ; and Ceawlin took many towns
and countless booty ; and, wrathful, he thence returned to
his own." The difficulty is to identify Fethanleag ; some
have suggested a place in Gloucestershire, in which case
the entry would be irrelevant here ; but Dr. Guest argued
for its identity with a place now called Faddiley, near
Nantwich, in Cheshire. In that case Ceawlin, marching
up the Severn valley, could hardly avoid having to do with
the people of the Wrekin district : he could not have ven-
tured further north without getting possession at least of
Viroconium, or of effecting its destruction, that is to say if
its destruction had not happened some time or other
previously.

This is, however, not a very satisfactory way of
trying to date a phonological change, so I would now
turn to Bede. It has already been suggested that the
Meilochon in his Ecclesiastical History seems to imply that
the name had, in Brythonic pronunciation, been modified
from Mailocon into Mailogon. But the same work contains
other names in point, such as that of Caedmon, the first
Northumbrian poet. He died in 680, and his name is a
form of that which Welshmen went on writing for a long
time afterwards as Gatman, now Gadfan. Similarly with
Gaedualla, both as the name of the Venodotian king, called
in Welsh Gatguollaun, later Gadwallon, who was blockaded
in the Isle of Glannog, or Priestholme, by the English in
629, and as the name of a West Saxon king who, according
to Bede, gave up his throne in 689, The early Celticform
of the name must have been Gatwuellaunos, the plural of
which is attested as the name of the Catuvellauni, one of the
most powerful tribes in Britain in the time of Caesar. Bede

32 All around the Wrekin.

mentions, also, a Welsh king Cerdic : his words are "sub
rege Brettonum Cerdice", and Mr. Plummer, the editor of
Bede's historical works, rightly suggests that this was
probably the Ceretic whose death is given in the Annales
Gambrioe, a.d. 616. The same name occurs also in the shorter
spelling Gertie, given in the Mistoria Brittonum to the king
of Elmet, expelled by Edwin of Northumbria. That is,
there were two Brytlionic forms, Geretic and Gertie, parallel
to such pairs as Dinogat and Bi/ngat, Tudawal and Tudwal ;
and the shorter form Gertie had reached Bede, with the t
reduced in pronunciation to d; so he wrote Gerdde.^

Here it may be asked, what about the unmutated c in
this name ; but the rule as to vowel-flanked consonants
does not apply. Mr. Plummer kindly informs me that it
was Bede's habit to place the proper name in apposition
to the appellative accompanying it, which means here that
the ending e of Gerdice has to be regarded as the Latin
ablative case termination supplied by Bede, the name as he
got it being Gerdie. Now a final consonant was not sub-
ject to more than half the mutational inducement which
was exercised on a consonant not preceded only, but also
followed, by a vowel. As a matter of fact the consonant
proves to have resisted much longer, and this persistence
has left its impress on the spelling down to the late Middle
Ages : witness the final t and c (less often p) regularly re-
tained in the spelling usual, for instance, in the Mabinogion
in the Eed Book of Rergest. The same remarks apply to
Bede's "in silva Elmete": he had the name as Elmet,

1 See Plummer's Bede, i, 256 (book iv, 23), ii, 247, and the Historia
Brittonum, loc. cit., p. 206 ; see also p. 177, where Vortigern's inter-
preter's name is variously given as Ceretic and Cerdic. StiU more
remarkable is the debftt in the Saxon Chronicle, a.d. 495, of a prince
whose name Cerdic or Certio suggests intermarriage with Celts even
earlier than can be implied by the case of Caedwalla,

All around the Wrekin. 2)%

which in Welsh is now Mfed, in English Elmt, as the
name of a district containing the parish church of Cynwyl
Elfed, so called to distinguish it from Cynwjl Gaeo, both
in Carmarthenshire. It is this Elvet, probably, that I seem
to detect in the bilingual inscription at Trallwng, near
Brecon, where the Ogam version reads Gunacennivi Ilvveto
'the Grave or Place of Cunacenniu of Elvet': this shows
the Welsh reduction of Im to Iv, for Im would have per-
sisted had the word been purely Irish. The Latin version
of the inscription will be mentioned later. Elmet, Elfed
was possibly not a very uncommon pl^ce-name : Bede's
instance survives in 'Elmet Wood', near Leeds.

Bede gives a still simpler instance, loc. cit., i, 82, namely,
'Dinoot abbas', the abbot of Bangor, who met Augustine
in one of the first years of the seventh century. In later
Welsh the name was Bwnawt, now Dimod, being the Latin
Dondtus, borrowed and pronounced at the time to which
Bede refers, probably as Dunot, with u tending to the
unrounding characteristic of the pronunciation of Welsh
u. When exactly the mutation of Welsh final consonants
took place in our Welsh texts has not, as far as I know,
been carefully studied. It is relevant to mention that the
sister dialects of Welsh, namely, Cornish and Breton,
appear never to have carried this mutation through. If
one consult Le Gonidec's Dictionary of Breton, one finds,
for instance, such alternatives' as tat and iad corresponding
to Welsh tad 'father', bet and bed to Welsh byd 'world'.
So with many more, including words where Le Gonidec

^ I take the forms ending with the tenues to be the older, but the
rules as to the use of the two sets do not seem to have been
exhaustively studied. Professor Joseph Loth has kindly referred me
to an article in which he has touched on them : see the Annates de
Bretagne, xviii, 617, also x, 30, where one of his pupils has discussed
an aspect of the same question,

34 All around the Wrekin.

suggests no option, such as oam/ih 'a little lamb', Welsh
oemg ; troadeh 'having feet, having big feet', Welsh troediog
'having nimble feet, active on one's feet', which is the com-
mon meaning given the vford in Gwyned; hevelep 'equal,
similar', Welsh cyffelyb 'similar', partly of the same origin
as the Breton adjective. It is possible that we have some
instances in Welsh itself : they would be short-vowel mono-
syllables of which there is no lack in Welsh ; but most of
them, when examined, prove to be English loanwords.

The foregoing notes on the proper names, preserved by
Bede, suggest two questions : the first is, when did the
English become familiar with the Brythonic names which
he gives as Gaedmon, Gaedualla, and Gerdic-e : perhaps
Aehhercwrn-ig 'Abercorn' should be added to them : see
Bede, i, 12. The Annales GambrioB carry us, in the case
of Cerdic, probably back to 616. We do not know
for certain when Csedmon, and Csedwalla of Wessex were
born, but before they were called by those names, time
enough must be allowed to have elapsed for intermarriage
or other processes of race amalgamation to render it
possible for Brythonic names to have had a chance of
emerging among the conquerors. On the whole the open-
ing of the seventh century appears by no means too early
as the approximate date of the earliest acquaintance of
the English with those three names. If that should prove
tenable one might, roughly speaking, lay it down that the
mutation of vowel-flanked tenues was an accomplished
fact by the year 600. The absence of that mutation in
the name Wrekin and its congeners does not enable us to
fix on a very much earlier time for the change, at most,
perhaps, half a century : so let us say 650, or thereabouts.
Nevertheless, the subtle and imperceptible beginnings of
the tendency to mutate the consonants, to slacken the
contacts made in pronouncing them, must date earlier,

All around the Wrekin. 35

since the same mutation system is characteristic of all the
Brythonic dialects.

The other question is, when did the mutation of final
tenues take place in Ceredic, Dunaut, Elmet, and similar
vocables. It will be found on enquiry that the tendency
to make that change had probably exhausted itself before
the period when the mass of English loanwords in colloquial
Welsh found their way into "Wales j for in them this
mutation is seldom found carried through. The following
may seiTe as instances, to which many more might be
added : adargop or adyrgoh ' a spider', a word in use in the
Vale of Clwyd, and derived from Old English attereoppe 'a
spider', also Welsh copa, cop or coh from coppe 'a spider':
the more common term for spider is in Welsh eopyn or pryf-
copyn. Another instance is clwt 'a rag or clout', from
some English form other than clout, which, in the sense of
a blow, has yielded the Welsh clewt 'a box on the ear';
and, lastly, Uac, from English slack, the meaning of which
it retains; whop 'a blow, stroke, or slap' (D. ab Gwilym,
poem 196), more frequently used as an adverb meaning
'with the suddenness or quickness of a blow', pronounced
in Cardiganshire whap, and in Glamorgan wap, while the
verbal noun in the former county is wahio 'to beat'. The
origin is to be sought in the dialectal English whap, wap
'to strike sharply or with a swing ; a blow, a knock, a
smart stroke': see Wright's English Dialect Dictionary.

There remains to be mentioned one of the most common
words in South Wales (except North Cardiganshire), one
that has always struck me as not of Welsh origin : it is
the word crwt 'a lad, a small boy', with its derivatives
crwiyn of the same meaning, and the feminine croten 'a
lass, a little girl'. To recognize the origin of these words
one has only to turn to Wright's English Dialect Dictionary,
and, in its proper place, one finds the word crut explained

d2

36

All around the Wrekin.

as meaning "a dwarf ; a boy or girl, stunted in growth".
The word is there stated to belong to Northumber-
land, Yorkshire, and Pembrokeshire, and the reader is
referred further to cvit and croot. Of these, trii is
explained as having, among other meanings, those of 'the
smallest of a litter' and 'a small-sized person', while
croot is given as the form usual in Scotland, meaning 'a
puny, feeble child ; the youngest bird of a brood ; the
smallest pig of a litter'. All this raises the question when
and whence crwi was introduced into Welsh: it looks as
though it was from Little England below Wales. When,
in that case, one bears in mind the former hostility
between Wales and that isolated England, it will not sur-
prise one that the word is not admitted into Welsh prose.
Similar questions attach to most examples of this class,
and few of them are regarded as literary words to be found
in Welsh dictionaries. An exhaustive and carefully classi-
fied list of them is much wanted. When made it would
probably throw much needed light on the intercourse
between the Welsh and the English from the time of
King Alfred down. An excellent beginning was made
some years ago, in his own dialect, by Prof. Thomas Powel
in the Gymmrodor; but search requires to be made in all
the Welsh dialects, as they have not always borrowed the
same words. This would form a good subject for research
work by one or more of the scholars trained by the
professors of Celtic at our University Colleges in the
Principality.

Reference has been made to the bilingual inscription on
a sepulchral stone at Trail wng, near Brecon : the Latin
version has been misread by me, and, T believe, by others.
What I make of it now, on the strength of a photograph
given me by the late Mr. Eomilly Allen, is the following : —

All around the Wrekin. '^'J

CVNOCENNI FrLIV[S?]
CVNOqENI HIC lACIT

That is to say : "The grave or the cross of Cunocenn : the
son of Cunogeii lies here." In the Ogam the equivalent
for Cuno-cermi is Oima-cennivi, and one perceives that there
was here a decided wish to keep to family names with the
same initial element Owno-, Goidelic Guna-, which has
already occupied us. In other terms, the two names
GvMocenn and Gunogen have to be carefully distinguished :
the former became in Welsh Goncenn (Goncen) or Gincenn,
and later Gyngen, pronounced Gyng-gen, while the latter
became successively Gongen, Gingen, with a soft spirant, gh,
which might either become i or else disappear. In the
former case we might expect Ginyen, which I have not
met with, and in the other Ginen, which would have, how-
ever, to be written Ginnen, as the first vowel remained a
blocked one and the later pronunciation and spelling were
Gyn-nen,'' not Gy-nen. The Booh of Llan Ddv^ carefully
distinguishes Goncenn from Gongen, as in the names of the
three abbots : "Concen abbas Carbani uallis, Congen abbas
Ilduti, Sulgen abbas Docguinni." Substantially this is also
the case with the oldest MS. of the Amiales Gcmhrice, and
with the Nennian Genealogies, both published (from the
British Museum MS., Earleian 3,859) by Mr. Phillimore in
the 9th volume of the Gymmrodor. There they are Gincenn
(or Gineen) and Ginnen, but some of the later MSS. of the
Annales Gawhriae, by retaining the g, which had ceased to
be heard, and writing Gyngen or Kengen (for Kennen),
appear to have misled not only Williams Ab Ithel, but
even more recent writers. The personal name enters into

' It is possible that Cennen is a variant of this name, to wit, in
Carreg Cennen, 'Cennen's Rock', on the top of which the ancient
Carmarthenshire castle of Carreg Cennen stands. At the foot of
that remarkable site flows the river Cennen.

^ See pp. 152, 154, 155, and others duly given in the Index.

^8 All around the Wrekin.

that of a farm called Cyneinog and Gyneiniog at the top
of the basin of the Eleri in North Cardiganshire. It
analyses itself into Cyn-ein-i-og = Owno-gen-i-aco^n,, and
compares with Bhufoniog from Bhufawn, Bhufon, 'Eoman
-us', PeuUniog from Poulin, Peulin, 'Paulinus', and
Anhumyawc, Anhmviog from Anhun 'Antonius'.

The Ounocenwi of the Latin of the Trallwng bilingual
has corresponding to it Gunacennivi in Goidelic, and from
Dunloe, in Kerry, we have a related form Gunacemi, where
the final a is all that remains of a genitive ending which
was probably ias. Later in the language one meets with
a feminine Gonchenn or Gonchend, genitive Gonchinni or
Gonchmne : the masculine also occurs, to wit, as Gonchend
or Goinchemi, genitive Goinchinn or Gonchind,^ correspond-
ing exactly to Gunocenn-d, Welsh Goncenn (Goncen), Cincenn,
Gyngen. The element cuno, Groidelic c^ma, in these names
has already been discussed, and the question remains what
we are to make of the other, cenno, Groidelic cenna. I am
now disposed to regard it as representing an earlier quenno,
Irish cenn, ceann, Welsh penn, pen, ' head or top, the end in
any direction'. We have another — probably an earlier —
instance of simplifying a medial qu into c, namely, in the
Carmarthenshire bilingual, which has Voteporigis in Latin
for Votecorigas in Groidelic. If this conjecture proves
admissible we can equate Gunocenn- with the GaUo-Roman
Gunopenn-ns, cited by Holder from Brescia, in North Italy,
G. I. L., V, 4216. The name would mean 'dogheaded', or
more probably, 'a head who is a dog', that is to say, dog
in the sense of a champion or protector, as usual in Celtic
names of this kind.''

Historically, the most important bearer of the name

1 See the Sev. Celtique, xiii, 290 ; 6 Huidhrin, note 597 to p. 109 ;
Book ofLeinster, flf. 325'' 325''' 326»- 351*-

2 See the Archceologia Camhrensis for 1895, pp. 307-13; 1907,pp.85-9.

J, Percy Clarkcy Llangollen.

Pillar of Ellseg, shewing the Concenn Inscription.

All around the Wrekin. 39

Concenn or Cincenn was one mentioned in the Nennian
Genealogies in the British Museum MS., Harley 3859 :
see PhiUimore's Pedigree xxvij {Oynwnrodor, ix, 181), where
he is called Cincen, son of Catel, also spelt Catell, later
Cadell. This latter is probably to be identified with
Cadell king of Powys, mentioned as Catell Pouis in the
Annales Gambrice, which record his death under the year
808, while the names of two sons of his occur under the
year 814, Griphiud and Elized. Now a monument of
capital importance, known as the Pillar of Elisseg, was
erected by Concenn in the neighbourhood of Valle Crucis
Abbey, n6t far from Llangollen. The Pillar had been
broken and fragments of it had been lost some time or
other before the inscription was examined in 1696 by our
great antiquary and philologist, Edward Llwyd. In a letter
written that year he sent a facsimile of what remained of
the writing to a friend, the letter and the copy are now. in
the Harleian collection in a volume which is alphabetical
and numbered 3,780. Since 1696 what Llwyd was able to
read has 'become nearly all illegible : so it has been deemed
expedient to have a photograph of Llwyd's copy submitted :
see pages 40, 41. This was rendered all the more necessary
owing to the astounding carelessness with which Gough,
Westwood, and Hubner have treated Llwyd's texb; but I
cannot go into details at present, as this paper has already
grown much longer than was intended.' It should be

' Gough printed both Llwyd's letter and his text in his Camden's
Britannia (London, 1789), vol. ii, 682, 583, plate xxii. The letter
was printed also in the Cambro-Briton in 1820, pp. 55, 56, and
recently a copy of it has been included in Mr. Edward Owen's
Catalogue of the MSS. relating to Wales in the British Museum, part ii,
410. That part, even more than the previous one, reflects great
credit both on the compiler and those who have the direction of
the Cymmrodorion Record Series. The letter is reproduced for
reference' sake at the end of this paper.

(2) FirtufrSROhcmccaBRohc/TiecLF/iivr

(3) eLire5 ELrre-5 FiLnir-Biio/uGCqc

(5) EdJFicauir Ai/wc iapidem proccvo

(6) roo eure^ -^iprc err Ciircs 9V' 'Jecr.

(7) ^^hepedrca'ce/npoi/op.ipc.-'iTjoR-c

(8) CcC^Em p€R y/m- ' € Po-iEfrrec^ECcnsto

(lo) _>— — . ./mqtiE RecfeUERi-c tn aiiercp-r-p

->=j7i c|ec ^e^cjic-criopfem rqpe
— ^-n eLire-sHHiPrEcrccoMce^/'

(12)

(13)
(H)
(IS)

.jgv ^ — ^'^1 ''^r=.a.lpl^ca.^irl

[40]

(ly) , — ' — — ' — — ■'Ji'n~€m~-fnotivan

(i8) -^ "'"^ """^

(19) — iLs—hc--ifnoftcLf\chicx.m

(20) — ecu mctximiir Sni-cc-ccpinaie

^2j) //// pccjxep — mccqncciiiiccii

(23) _.(|^£6£/'£d-"5ERmsf^yrq76

(24) — =pep£r;'s €ire-mce ^a/GcmOCXimi

(25) — sirqyjoccicfrc/^eBew Ro/napo

(26) /^qm'^cojJmccRchPipxi'c^ioc

(2;) c/i 1 ROB h^fF ae-se r^5© porcep-ce

(28) concep^^Se/ieJicciodfTi/Rcop

(29) cePH ecirnT'co-cccf=am!Lis3ji|r

(30) g-^ /I T:o"ca Eo&s/o/Je poiroir

(31) ypcjqeip

[41]

42 All around the Wrekin.

mentioned that Llwyd some ten or eleven years later
endeavoured to give in printed characters a facsimile of
lines 23-28 of the inscription. They are to be found in
his ArchoBologia Britannica (Oxford, 1707), i, p. 229", where
he uses among other letters a Greek /* for N, and several
letter-forms now used only in writing Irish. Put into
ordinary English letters, the lines in question run as
follows, differing slightly from the copy in 1696, which
has here been submitted in photography : —

.... bened .... Germ anus que

.... peperit ei se . . ira filia Maximi

regis qui occidit regem Romano

rum * Conmarch pinxit hoc

chirograf u rege suo poscente

Concenn ^ &c.

The Llwyd copy, reduced to what is intelligible at a
glance, but extended by the insertion of individual words
suggested by the context, and of certain formulae of a well-
known description, will stand somewhat as follows : —

(1) fConcenn fiUus Cattell CatteU (i)

(2) Alius Brohcmail Brohcma[i]l filius

(3) Elise^ Eliseg filius Guoillauc

(4) fConcenn itaque pronepos Eliseg (ii)

(5) edificavit hunc lapidem proavo

(6) suo Eliseg f Ipse est Eliseg qui (iii)

(7) .... hereditatem Pouo[i]s

(8) ... per viiii' [anwos] e potestate Anglo-

' After I had made repeated attempts to understand the text, my
friend Professor Sayce kindly came to my assistance, and he has
carried the interpi'etation further than I could. Thus, for instance,
at the end of line 6 and the beginning of line 7 he would read nacttis
erat ; and here, I believe, I owe to him the reading viiii, for
Llwyd's dots seem only to suggest vim. Before leaving for the
Soudan he gave me to understand that his emendations would be

All around the Wrekin. 43

[rum] in gladio suo parta in igne
[tQuicjumque recit[a]verit manescrip- (iv)

[turn lapid]eu] det benedictionem supe-
[r anima] in Eliseg f Ipse est Concenn (v)

manu
ad regnum svum Pouo[i]s
et quod

montem
(One line wanting, perhaps more) (vi?)

monarchiam
Maximus Brittanniae
[Conce]nn Pascen[t . . . . ] Maun Annan
[ tJBritu a[u]t[e]m filius Guarthi (vij)

[read Guorthi]

23) [girn] quern bened[ixit] Germanus quem-

24) [qu]e peperit ei Se[v]ira filia Maximi

25) [re]gis qui occidit regem Eomano-

26) rum t Oonmarch pinxit hoc (viij)

27) chii-ografum rege suo poscente

28) Concenn f Benedictio dojwini in Con- (ix)

29) cenn et svos in tota[m] familia[m] eius

30) et inw tota eagionem [reati in totam earn

regionem] povois
(31) usque in [diem iudicij

To check the lacunae, more or less, we have Llwyd's
spacings, but they cannot be relied on so much as the
number of letters to the line. TJp to line 25 inclusive, the
lines that permit of being counted make an average
exceeding 28 letters a line. From line 25 onwards the

published in the Arehceologia Cambrensis as part of his address to the
Monmouth meeting of the Cambrians in September last. The
October number has been issued, but does not contain the account
of that meeting : it will probably be in the January part.

44 All around the Wrekin.

inscriber has taken more room, and the average falls to 24,
The whole inscription was divided into paragraphs, with a
cross placed at the beginning of each. The third of the
paragraphs begins with Ifse, estEUseg qvd, etc., a very Celtic
construction, meaning 'It is Eliseg who' did so and so.
The paragraph seems to relate how Eliseg added to his
dominions by wresting from the power of the English a
territory which he made into a sword-land of his own, 'in
gladio' suo'.

Paragraph v is mostly hopeless, but it seems to
summarize the achievements of Concenn himself, especially
as regards the additions which he made to his realm of
Powys. Then followed probably a paragraph stating that
Eliseg's mother was Sanant, daughter of Nougoy (or Noe),
descended from Maximus (Ped^ ii and xv), and closing with
a sentence giving the names of five sons of Maximus. I am
not clear how the sentence ran, but possibly thus : — "Prius-
quam enim monarchiam obtinuit Maximus Brittannise,
Concenn, Pascent, Dimet, Maun, Annan genuit." Goncenn
is a mere guess : perhaps Maucann would be better, but
any name in nn is admissible. Bimet, which in the
Pembrokeshire bilingual inscription at Trefgarn Each -is
Demet-i, seems to fit the lacuna, and a bearer of that name

1 The full term in Irish appears to have been 'to clean or clear a
sword-land', or 'to make a land of the sword' of it. The land itself
was called claideb-thir or tir claidib, which came to be called simply
claideb or cladeom 'sword'. Possibly in the case of the two Pembroke-
shire rivers Cleddau 'sword', the word originally meant the districts
drained by them, and seized by the Dfissi as their sword-lands in
Dyfed. See Celtic Britain, p. 196, Skene's Chronicles of the Picts and
the Scots, pp. 10, 319, 329, and the Book of Leinster, f. 383»' SSS"'
Compare also Meyer's " Expulsion of the Ddssi " in the Cymmrodor,
xiv, 116, 117, where we meet with the phrase do aurglanad rempu
'to clear (the land) before them' of its inhabitants. In igne, mean-
ing 'with Are, by means of fire', is a literal rendering from Celtic: see
the same story, pp. 114, 115.

All around the Wrekin. 45

is mentioned as a son of Maximus in Pedigree ii, which
makes Dimet an ancestor of Coneenn through Bliseg's
mother Sanant. Maximus is said to have been a native of
Spain, but Dimet's name is of importance as indicating a
connection between Maximus and Dyfed, the country of
the ancient Demetse, perhaps through his supposed British
wife, the Elen Liiydog of Welsh legend. Add to this the
fact of that legend associating him with Caerleon and
Carmarthen, and, above all^ calling a Dyfed mountain top'
after him Gadeir Vaxen 'Maxen or Maxim's seat'. Annan is
probably to be corrected into Annun, given as Anthun son
of Maximus in Ped. iv. It is the Latin Antonius, with the
nt reduced into nn as in Maucann, by the side of Maucant
ii) Ped^ xxii and xxvii : it is otherwise spelt Annhun or
Anhun as already mentioned. The MS., Jesus College xx,
gives Maximus {Cymmrodor, viii, 84, 86, 87) three other sons
all with their names derived from Latin Owein, older spelling
Eugein = Bugenius, Gustennin = Gonstantinus, and Dunadt
= Ddndtus.

The next paragraph runs as follows, beginning in a
Celtic fashion without a copula: — "Britu autem filius
Guorthigirn, quem benedixit Germanus quemque peperit
ei Severa filia Maximi regis qui occidit regem Romanorum."
For 8evira is doubtless a spelling of Severa, but whether a
daughter of Maximus of that name is mentioned anywhere
else I cannot say. To put this important statement right

1 See 'Maxen's Dream' in the Oxford MaMnogion, p. 89 : the
Pedigrees give the name as Maxim, but even that is not really
ancient : the old form would have been Maisiv, later Maesyf, which
must be supposed superseded by the book form Maxim. It is a
difficulty ; and there is another, namely, how Maxen came to supersede
Maxim. The former recalls Maxentius, without, however, being
correctly derived from that name. Mr. Wade-Evans, in the Cymmrodor
xix, 44, note 4, suggests that our man was a Maxentius, and not the
Maximus who became emperor in the West.

46 All around the Wrekin.

with the Nennian Pedigrees, the latter have first to be
corrected in certain particulars. One of the foremost
things to attract one's attention is the fact that they never'
mention Gruortheyrn or Vortigern. For his name they
substitute "Cattegirn, son of Catell Durnluc" : this seems
done partly for the sake of CateU or Cadell, the pet
convert in the story of St. Germanus's miracles as given in
the Historia Brittonwm, he. cit., p. 176. There the Saint
is made to tell Cadell, one of the servants of Benlli, that
he, Cadell, would be king, and that there would always be
a king of his seed. The story proceeds to exaggerate the
prophecy as follows : — " Juxta verba Sancti Germani rex
de servo factus est, et omnes filii eius reges facti suat, et a
semine illorum omnis regio Povisorum regitur usque in
hodiernum diem." So the Nennian Pedigree xxii ends
with "map Pascent | map Cattegirn | map Catel dunlurc",
though the Fernmail Pedigree in the Historia Brittonum,
loc, cit., p. 193, has "filii Pascent filii Guorthigirn Guor-
theneu", without a trace in any of the MSS. of either
Cattegirn or of Catell. Pedigree xxvii, however, emphasises
Ped. xxii, as it ends with "map Pascent | map Cattegir[n] |
map Catel | map Selemiaun". Here the father of Cadell
seems to have been an unnamed man belonging to Cantrev
Selyv, in Brecknockshire. This looks ingenious on the
part of the scribe, as Cadell was described in the Germanus
legend as rex de servo factus. The difficulty is avoided in the
MS., Jesus College xx {Gymm., viii, 86), where we have words
to the following efPect: — Cassanauth Wledig's wife was
Thewer, daughter of Bredoe, son of Kadell deernlluc, son

1 In studying these pedigrees I have found Mr. Phillimore's edition
of them in the Cymmrodor, vol. ix, invaluable, and next to that
Mr. Anscombe's "Indexes to Old Welsh Genealogies" in Stokes «&
Meyer's Archiv fiir celt. Lexikographie, i, 187-212. See also p. 514,
where he has anticipated me as 1;o Severa.

All arotmd the Wrekin. 47

of Cedehern (= Cattegirn), son of Gwrtheyrn Gwrtheneu.
This makes Cadell grandson of Gwrtheyrn or Vortigern.
The Bredoe of this pedigree I take to be the same name as
Brittu in the Nennian Ped. xxiii, which ends with "map
Brittu' I map Cattegirn | map Catell". Making here the
correction found necessary in the other cases we get
"map Brittu | map Guorthegirn". That this hits the mark
is proved to a demonstration by the "Britu autem filius
Guarthigirn" of the Elisseg Pillar.

If we try to look now at the inscription as a whole we
perceive that the object which Concenn had in view was
the glorification of himself and Eliseg (1) on the score of
their own achievements, and (2) by reference to their
ancestors, the Emperor Maximus and the King Gwrtheyrn
or Vortigern. The Powys dynasty was Goidelic, and prob-
ably the "Welsh epithet in Gwrtheyrn Gwrtheneu, which
Williams ab Ithel, at the beginning of his edition of Brut
y Tyivysogion, has rendered into English as 'Vortigern of
Eepulsive Lips', simply meant that Gwrtheyrn spoke a
language which was not intelligible to his Brythonic
subjects, or at least that he spoke their language badly.
Here one cannot help realizing that the inhabitants of
what is now Wales could not then have had any collective
name meaning men of the same blood or men who spoke
the same language. They could hardly adopt any name
in common, which was not comparatively colourless. So
there eventually became current an early form of the word
Oymry, which only meant dwellers in the same country.
In fact Cymry connotes the composite origin of our Welsh
nationality. By the beginning of the ninth century,
however, the dynasty had practically become Welsh,

^ The name occurs in one of the Tomb Verses, no. 36, in lit/d
Britu 'Britu's Ford', so the modern prominciation should probably
be Ehyd Bridw,

48 All around the Wrekin.

which possibly made it all the more necessary in the
opinion of Concenn and his Court to place on record what
they considered a true account of Gwrtheyrn's position
with regard to Maximus and to St. Germanus, as con-
trasted with the ugly stories which the Brythons associated
with his name. There is, therefore, no hope of reconciling
the testimony of the Pillar of Elisseg with the legends in
the Historia Brittonum in so far as they concern Gwr-
theyrn's character.

The Historia, however, throws a ray of light on
Gwrtheyrn's origin; for in Fernmail's pedigree he is
said in two of the MSS., one in the Vatican and
the other in Paris, to have been the son of Guitaul,
son of Guitolion or Guttolion;^ but those names are
simply the Welsh adaptations of the Latin Vitalis and
Vitalianws. Most of the MSS., it is true, have instead of
Guitolion the form Guitolin, but this was a different though
kindred name derived from the distinct Latin name
Yitalinus, In fact Guitolin occurs later in the Historia
Brittonum, namely, in sec. 66. Most of the scribes have,

^ See the readings given in Mommsen's edition, loc. cit., § 49
(p. 193), § 66 (p. 209); and for his account of the MSS. see pp. 119-
21. The Vatican MS. was published by Uunn (London, 1819): for
its reading of the Fernmail pedigree see p. 78. It is remark-
able for combining such old spellings as JEmbres and Tebi with
such a comparatively late form as Teudor, in Mommsen's text
Embreis, Teibi, TeudvMr respectively. The first element in this last
name is tew 'thick', used probably with the force of 'very, exceedingly',
and the second, dvjnr, became successively dwfr, dior, so the later
form of the name is Tewdwr. Compare Welsh dvbr, dwfr 'water',
which in colloquial Welsh is always dwr. The meaning, however,
of dubir, dwr in the personal name has to be guessed from the
probable equivalents in other languages, such as English, where it is
dapper, Modern German tapfer 'valiant'. Old Slavonic dobr& 'beautiful,
fine, good'. Some would also connect the Latin faber 'smith' as
meaning the man of a cunning art or craft. So Tewdwr may have
signified 'very good, very fine, very clever', or possibly 'very vaUant',

All around the Wre'kin, 49

not unnaturally, made ChittoUon or Gwitolion into Gfuitolin,
except the two which I have specified: for them the
temptation to reduce the name in -ion into Guitolin prob-
ably did not exist, as their texts do not appear to contain
sec. 66. IsTow the former name occurs on a bilingual
tombstone at Nevern, which reads in Ogam simply
Yitaliani, meaning 'the monument or place of Vitalianus
or Guttolion', and in Latin letters of the most ancient
type perhaps to be found in our non-Roman inscriptions : —

VITALIANI

EMERETO

This is so condensed that it is difficult to be sure of tlie
exact meaning, but it seems to suggest that the deceased
was regarded as holding some rank in the Roman army,
and the ease may be compared with the later Dyfed
bilingual from Castell Dwyran,' where the deceased has the
Roman title given him of 'protector'. Such cases help to
answer the question how it was that during the later years
of the Roman occupation the troops of whom we read were
all in the north and east of the Province; for it would
seem that the west was to be looked after by the chiefs of
the Dessi. The latter, on the other hand, appear to have
pursued a more or less romanizing policy, as may be
gathered from the Latin names to be found in Goidelic
inscriptions both in Wales and Ireland, such, in the
former, as Pompeius and Turpilius, Severus and Seuerinus,
and, in the latter, such as the VitaUnus already mentioned.
For besides the D^ssi who came over to Dyfed, there were
others who coasted westwards and landed in Kerry. It is
to them, probably, one has to refer an Ogam inscription
including the name Vitalin, found at Ballinvoher, in the

' See ArchcBologia Cambrensis, 1895, pp. 307-13, and the Cymmrodor,
vol. xviii, = 'The Englyn', pp. 72-4.

50 All around the Wrekin.

barony of Corkaguiny in that county. At a well near
Stradbally, in co. Waterford, the land, to this day, of the
Dessi, I have seen an inscription involving the genitive
AgracoUn-i, which T take to be a derivative from Agricola.
The motive here was doubtless admiration for the fame of
the great Roman general of that name. In the case of a
group like Vitalis, Vitalianus, and Vitalinus, the motive was
different but not far to seek : the names were chosen as
involving mta 'life', probably by a family whose Goidelic
names began with an early form of the vocable beo, in
Welsh hyw 'alive, quick', such as Beodn, Beoe, Beo-aed,
Beo-gjM, which was borrowed into Welsh early, and
modified eventually into Beu-gyio, Beuno. Time would fail
me to do justice to all the conclusions to be drawn from
the facts to which I have called attention. There is one,
however, on which I wish to lay stress, and it is this : the
Vitalianus stone at Nevern probably marked the grave of
the grandfather of Gwrtheyrn, son-in-law of the Emperor
Maximus.

VI.

To return to the Pillar of Elisseg, it has always struck
me that it is a column obtained from some Eoman building
of respectable dimensions ; but where? The inscription
upon it must, when perfect, have formed a historical
document, with which we have absolutely nothing of the
same importance to compare. There remains one thing to
be done to lessen our loss from the treatment to which the
stone had been submitted before Ed. Llwyd's examination
of it, and that is to have a thorough search made for the
missing fragments. Regardless of expense the little
mound, on which has been set up what remains of the
original pillar, should be carefully sifted, and the hedges
near should be ransacked until the broken pieces have

All around the Wrekin. 51

been found. In any case they cannot be far away, and
they have probably escaped the weathering which has
reduced almost to illegibility the exposed portions of the
pillar. Let us hope that some generous Cymmrodor will
come forward to help us in the search which I have sug-
gested. It is also highly desirable that good' casts should
be made of the pillar as it is and before ^ it has become
completely illegible.

The fact that Concenn, king of Powys about the
beginning of the ninth century, bore an Irish name, has,
as far as I know, never been detected, and still less, if
possible, that his great-grandfather Eliseg's name was also
Irish. So I have to dwell a little on the latter : Edward
Llwyd has copied it as Eliseg the five times which it
occurs in the inscription ; but in the Genealogies it is
usually Elised, as also in the Annates Gambrice, a.d. 814,

943, 946. On the other hand the Liber Landavensis
regularly spells it Mised, and so with the Latin genitive
Elised-i in the Book of St. Chad ; but a form Elisse also
occurs, as, for instance, in Brut y Tywysogion, a.d. 815,

944, while under 1202, in the same, we have it twice as
Elisy.^ These, without the final d, practically prove the
consonant to have been sounded as the soft spirant d or
dd, a sound wliieh was sometimes represented in Old Welsh
by t. Hence the final t of Mitet in Pedigree xxvij (p. 181) :
the other t of that spelling was probably a result of the
scribe misreading a or a reversed s as t.' Thus the older
spellings in Welsh practically reduce themselves to three,

g, Mised, and Elized. The Irish name occurs in a

' Possibly Elisei, which occurs once as the name of a witness in the
Liber Landavensis, p. 216, is to be regarded as an instance of this
name.

^ How this can have happened may be seen from the way in which
Oriali or Crisdi in a Margam Abbey inscription used to be read
Critdi: see the Archeeologia Cambrensis, 1899, p. 142.

E 2

52 All around the Wrekin.

genealogy of the Dessi in the Book of Leinster, fo. 328'', as
Heslesach. The man so named stands twelfth in descent
from Artcorb, whose son Eochaid was leader of those of the
Dessi who took possession of a part of Dyfed about 265-70.
The initial aspirate forms no etymological part of the
name ; so the more regular spelling was doubtless Eslesach,
which would be that of the nominatiye. The genitive
should be Eslesaig, and it occurs in the same MS., fo. 340",
spelt Eislesaig, where the apex means that the pronuncia-
tion of esl had been modified in actual speech into el.
Welsh made si into stl, while Irish reduced it into I or II,
with or without vowel compensation. Thus Welsh gwystl
'a hostage' is in Irish giall, of the same origin as German
geisel, Old H. German gisal : in fact, the German was
probably a loan from some Celtic language of the Con-
tinent. Or take the Welsh name Ygeestyl, Engistil, the
Irish, equivalent of which is found written in Irish,
Ingcel and IngelU The pronunciation of the g at the
end of a genitive of this kind was that of a very evanescent
palatal gh, and the retention of the g of Miseg was
historical rather than phonetic. But the Irish sooner or
later treated every dh as if it had been gh ; and Irish gh,
influenced by the vowel i or e, passed into the semivowel or
consonant, i or y,' which Welsh pronunciation had once a
habit of converting into tf, now written dd, as for instance
in Iweryd (for lueriiu), Iwerdon (for luerion-os), Irish Eriu
genitive Erenn, 'Ireland'.

It remains to say something about the spelling with
z, a letter which looks equally singular in Welsh and
in Irish, for neither language has the soft sibilant in

1 For more instances see Rhys's Celtic Heathendom, p. 567, and
Celtic Folklore, p. 542 ; also Archaologia Cambrensis, 1898, pp. 61-3.

2 See my Manx Phonology, pp. 118-23; and as to Welsh tf f rem
i or y, my British Academy paper Celtts S; Galli, p. 13, note.

All around the Wrekin. 53

its pronunciation. But in Medieval Irish z was treated
as an orthographic equivalent for sd or st; so we have in
the later portion of the Book of Leinster, ff. 357", 357'',
358'', SSS"!, 364^ Zepliani for Stephani, and ff. 341, 353%
364'', Zrafain for what is there otherwise written Srafain
and Srafddn, nominative Srafan, seemingly for an earlier
Strafan: Stokes, in his Martyrology of Gorman, p. 397,
cites 8trofan from the Martyrology of Tamlacht. Vice
versa we have Elisdabet^ for Elizabeth, and Steferus' for
Zephyrus. More illuminating, however, is the name of an
Irish bishop given in the Martyrology of Oengus as Nazair,
July 12, and p. 168. It occurs also in the Booh of Leinster,
ff. 812", 315% 335'', 348', 351", 351', as Nazair, both
nominative and genitive, but the genitive of what appears
to be the same name occurs, fo. 337^, as Nadsir. This
suggests that the name is to be regarded as syntactically
made up of Nad-sdir, with nad as the unaccented form of
nioth 'nephew', and sder 'artificer'. In that case the z of
Nazair represents here, not sd or st, but ds or ts, and the
origin of the spelling with the z becomes clear at a glance.
It is to be sought in such Greek spellings as SSeu? for
Zeu9, and the like, and in the teaching of the old gram-
marians that 5 was pronounced ah or else So-.' In a Latin
list of bishops ordained by St. Patrick, one detects the
name Nazair made into Nazarius, and that form, coming,
as it does, from the Booh of Armagh, a MS. finished in the
year 807, carries the z back to the eighth century.*

^ Stokes's Martyrology of Oengus, p. 110, a propos of April 1.

2 O'Donovan's Battle of Magh Bath, p. 238.

^ In either combination the sibilant meant the sonant s which in
English and French is written a. See Georg Curtius's Erlduterungen
zu meiner griechischen Grammatik (Prague, 1870), pp. 17-19, and Blass,
fiber die Aussprache des Griechischen (Berlin, 1888), pp. 113-122.

' See Stokes's Patrick, p. 304, Stokes & Strachan's Thesaurus
Palteohibernicus, ii, 262, also pp. xiii-xv. One of the most singular

54 All around the Wrekin.

All this would seem to imply that the name was
Eslestach, when the spelling with » was first applied to
it : Irish reduces sd, st, ds, and ts all to ss or s, though how
early it happened in the case of sd, st, it is hard to say.
The name might in that case be regarded as a contraction
of some such a longer form as Eselestach, derived from
Eselest or Eseles. I suggest this because we have at the
top of Ped. xxiij, a name esselis, the initial letter of which,
like other initials in the Nennian Pedigrees^ the rubricator
neglected to insert. I guess it to have been an h to help
to make up Hesselis, which, with the accent on the first
syllable, would be liable to be contracted in. Irish to Eislis
or Eisles — there was an Irish name Aneisles, Aneislis —
whence probably our Welsh name Eli-s, spelt also Ellis with
English II. The only other name which the -esselis of the

things connected with the letter ^- in Irish is that one of the Ogam
symbols, not yet found in an ancient inscription, namely, the 14th, is,
in a tract on Ogttiic alphabets in the 14th century MS. of the Book
of Ballymote, named xmif, S. 309a' lines 21, 45; 309''' 1. 33; 310». 1. 40.
O'Donovan, in his Grammar, p. xxxii, treats this as straif, and inter-
prets it as "the sloe tree"; for it belongs to an alphabet which has the
individual symbols called by tree-names. From this arose the
untenable notion that the Ogam in question stood for st or z. The
sound originally meant was probably that of / or ph, a phonetic
reduction sometimes of Indo-European sp or sp'h. This / has since
been mostly changed into s, and the symbol is lost in favour of the
Ogam originally representing s. The change, into s took place
initially, while / still remained as a non-initial, and the man who
first called the/ Ogam «f)m/ could, doubtless, not find an instance of
its use as an initial, so the name straif may be regarded as aptly
chosen. In Irish, initial / stands, since the eighth century or
thereabouts, mostly for the provected sound of v or lo, and not for
an original / at all ; but among other instances of /, derived from
original sp, and still remaining / in Welsh (now written ff)t i^^V te
mentioned Irish seir 'a heel ', nominative dual dd seirith, but accusative
tria adipherid 'through his two heels' (Stokes's Celtio Declension, p. 26) :
the Welsh is ffer 'the ankle', Greek trct>vp6v, the same. See also
his Urlceltischer Sprachschatz, p. 299, where he cites 'bd tri .sine' 'of a
cow of three teats', otherwise ' bd triphne', where sine anA-phne are pro-

All around the Wrekin. 55

MS. could possibly suggest is what is .usually treated as
Ll&oelis or LlefeUn : this ought, doubtless, to be Llewelis
or Lleuelis, to be analysed Lleu-elis, As to this use of
Lieu compare Old Welsh Lou-hrit or Leu-brit to be equated
with Logu-qurit- in an Ogam inscription (in the Nat.
Museum, Dublin), later Luicrith : it would mean 'one
who has the form or countenance of Lieu or Lug'.

The five names in the first clause of the legend on the
Pillar of Elisseg^ are, as read by Ed. Llwyd, Concenn, Cattell
(wrongly Catteli), Brohcmail and Brohcmal, Eliseg, and
Gruoillauc. Of these Concenn and Eliseg have been shown
to be of Goidelic origin. Broccmail is a name common to
Brythonic and Goidelic, or else a loan from Goidelic : the
common Welsh spelling is Brochmael, and the Old Irish
would be Broccmdl, genitive Broccmail, but at present I

bably f ormii of the same origin as Anglo-Saxon spana 'teats or speani<\
Other names in the tract in the Book of Ballymote for the / Ogam
are the following, ff. SIO". Is. 34, 48; Sli"' 1. 4 :— (1) A place-name
Si-uthar, derived probably from sruth 'a stream', Welsh jfno(i, possibly
from the same root as German sprudel 'a well, a fountain'. (2) Sust,
which is the Latin word fmtis borrowed, as is the Welsh equivalent
ffust 'a flail'. (3) Sannan, a saint's name, probably identical with
Fanon-i in the Latin of a Devon bilingual, now in the British Museum.
Compare Fannuc-i from a Latin inscription in South Pembrokeshire,
which recalls Irish Sannuck, the name of one of St. Patrick's monks.
See Stokes's Patrick, pp. 305, 412, but take note of Sanuous, Sanucin-o,
C. I. L., V, 2080, XIII, 5258. (4) There are other names there of
which I know not what to make, such as Zur, that of a 'linn' or water,
hardly Siuii- 'the Suir', and Zeulce, the name of a dinn or height, and
zorcha 'light or bright'.

' Since this was written my attention has been drawn to the
pedigree of Cerdio in the Saxon Chronicle, a.d. 552, where one reads
that Cerdic was Elesing, that is, son of Blesa, and Elesa was Esling,
that is, son of Esla. Here there is not only a striking similarity
between Eliseg and Elesa, but two names, Elesa and Elsa, to compare
with the two Eliseg and Elis, or rather, with the Goidelic forms from
which they derive. Even were it to be urged that Elesa and Esla are
due to a meaningless duplication the residue of similarity is signiiieant.

56 All around the Wrekin.

cannot lay my finger on an instance. The Welsh Broch-
mael should regularly be pronounced Brochvael, or rather
Brychvael, but what has come down to us is Broehwel,^
which is a modification of the Irish genitive BroecmdU,
pronounced Brocwel with the accent on the first syllable,
accompanied with a shortening of the second. This leads
me to expect that Cattell or Gatel may prove to have been
Goidelic too : the name which in that case it represents
must have be^en the Irish Gathal, genitive Gathail, for an
early Gatyxil-i = Gatu-uaUi, in Welsh Gatwal, Gadwal.
Possibly it is in the name of some Irish Gathal that we
have to seek for the Gadwal after whose name the commot
of Gedweli or Gydweli was called : the English spelling is
now Kidwelly, with the accent on the second syllable
and II pronounced as in English. Somewhat similar
remarks might be made on Guoillauc, which occurs in
pedigree xxvii as Gwilauc.

Enough has now been said to shew that the Powys
dynasty of Eliseg was a Goidelic one, and I will only add
a mention of a passage in the MS., Jesus Gollege xx, § 23 ; see
the Gymmrodor, viij, 87, where the mothers of Einion and
Cadwallon Lawhir, the father of Maelgwn Gwyned, are
described as daughters to Didlet, king of Gwydyl Fichti in
Powys. Whether these were Goidels or Picts is not
certain, nor is there any indication where in Powys they
were located." The question suggests itself whether at

* My previous attempts to account for this form have been
unsatisfactory ; and for one or two other instances of the popular form
of a name in Wales being more Irish than Welsh see my Celtic
Folklore, pp. 641, 542. Compare the case of Docmael, Dogmael : two
of that saint's churches are called Han-Ddogwel and ' St. Dogwel's,'
and a third Llan Dydoch (= Do-Tocc-), in English 'St. Dogmael's',
retaining an old quasi-official spelling Doyniael. See Rice Rees's
Welsh Saints, p. '211.

^ Who were the five chiefs Wydyl Ffichti mentioned in the short
poem, xlix, in the Book of Taliessin (Skene, ii, 205) ? The number.

All around the Wrekin. 57

the outset the Goidels of Powys extended their power to
that region from the direction of Buallt and the Wye, or
from Gloucester and the Severn. On the one hand,
Fernmail, descended from Pascent sou of Gwrtheyrn, was
king of the Wye districts of Buallt and Gwrtheyrnion
about the end of the eighth and the beginning of the
ninth century.' On the other hand, legend associates a
branch of the Dessi with Caer Loyw' or Gloucester,
apparently the same branch which was descended from
Pascent son of Gwrtheyrn. In other words the ancestors
of the Eliseg family may have pushed northwards along the
Severn valley in the direction of Pengwern Amwythig and
Wales. All this, however, is merely touching the surface
of the history of the Dessi in Wales and the Marches, but
even so we have stumbled across some important data for
the writing of a new chapter on the most obscure period
of Welsh history. It only remains for me to mention one
or two subjects which it would be desirable to have studied
in connection with it. Such, among others, are the distri-
bution of Goidelic inscriptions in South Wales, the preva-
lence of Goidelic proper names in the diocese of LlandafE,
as attested by the lAher Landavensis, and the so-called
hreiniau or privileges of the Men of Powys.^ Finally,
should the evidence point to the conclusion that the Dessi
pushed their conquests up the vale of the- Severn, it
could not help suggesting at the same time the question,
whether it was not they that destroyed Viroconium.

five, suggests the men in the first clause of the Blisseg inscription,
though none of them can have been contemporary with Cadwallon
Lawhir's mother's father.

' See the Ristoria Brittonmn, loc. cit., p. 193, and Zimmer's
Nemdus Vindicatus, p. 71.

2 See my paper on " The Nine Witches of Gloucester", in the volume
of birthday essays, presented to E. B. Tylor (Oxford, 1907), pp. 285-93.

^ See the Myvyrian Archaiology, i, 257, and Aneurin Owen's Ancient
Laios and Institutes of Wales, ii, 742-7.

58 Ail around the Wrekin.

APPENDIX I.

Mb. Stevenson's Monograph on the name Weekin.
(See, f. 89 above.)

The earliest mention of the Wrekin occurs in the dating
clause of a charter of 855, derived from the late eleventh
century Worcester chartulary "quando fuerunt pagani in
Wreocensetun" {Cart. Sax., ii, p. 89). This is an older
name than Shropshire for the district about the Wrekin
(or, strictly speaking, the people of the Wrekin). They
are probably the Wocensoetna (gen. pi.) of the list of early
territorial names {Cart. Sax., i, p. 414) upon which Pro-
fessor Maitland has conferred the name of the Tribal
Hidage. This is derived from a tenth or eleventh century
MS., which contains many corruptions. A thirteenth cen-
tury copy {Ibid., p. 415) reads Porcenseteve (by confusion of
the O.E. sign for W with P, which it greatly resembled), so
that the original probably read Wroeen-soBtna. This form
occurs in another Winchester charter dated 963 (Ibid., iii,
355, from the twelfth century Godex Wintoniensis) "in pro-
vincia Wrocensetna".

The Wrekin itself is mentioned in a charter, derived
from the same chartulary of 975 {Ibid., iii, 650) "on
Wrocene", "andlang Wrocene" in boundaries near Up-
pington, CO. Salop. Here the name is, apparently, de-
clined as a feminine 6-stem, with a nom. sing. Wrocen and
a short vowel in the root syllable. The absence of the
demonstrative pronoun proves that Wrocene is the name
of some local feature and is not a common noun. Celtic
local names usually appear in the O.E. charters without
inflexion and without the demonstrative pronoun, as
pointed out by Professor Sievers in Paul and Braune's
Beitrage eur Geschichte der deutschen Sprache und Literatur,
ix, p. 251.

The Abingdon chartulary contains a charter of 944
{Ibid., ii, 557), which mentions in the boundaries of Blew-
bury, CO. Berks, "be eastan Wrocena stybbe Jjset swa to
Wrocena stybbe, Jjonne of Wrocena stybbe". In form
this seems to be a genitive plural, but no such word is
recorded in O.E. One would expect a tree-stump to be

AU arozmd the Wrekin. 59

known by a man's name or by an adjective or partici-
pial compound. This name is probably unconnected with
that of the Wrekin.

Apart from this last instance, we have evidence that
the name fluctuated between Wreocen and Wrocen. The
instances are too numerous to be ascribed to clerical errors,
and it is evident that the two forms existed both in the
name of the Wrekin and in the local names formed from it.
Professor Napier suggests that the Wrocen form arises from
Wreocen through labialisatiou of the r produced by the
initial W. The variation seems to be clearly due to phonetic
action, and not to arise from different forms originally.

In this case we may regard Wreocen as the original
form. This may be explained as a Mercian development
(with the change of e or i to e.%, iu, later eo, produced by a
following u) from an original Wrekun or Wrikun. The
latter would have been the form necessarily assumed in
O.E. by an early Celtic Wrikon-.

From the evidence of the forms it is obvious that
Wreocen was exempt for dialectal or other reasons from
the Anglian "smoothing" before c, by which Wreocen
should have become Wrecen. The modern form of the
name descends from Wreocen. The Wrocen forms seem to
shew that the diphthong was sometimes accented on the
second vowel.

Wrocwardine, Salop, represents an O.E. Wreocen-
weor^ign (the latter part of the compound usually becomes
-wardine in local names in this district; it is related to
weorS, weor^ig 'village, farm'). It appears in Domesday
several times as Recordin(e), where the Norman scribe has
not represented the initial w of the O.E. form, as is
usually done in the Survey. But the Rec- represents
regularly, with the exception of the suppression of the
initial consonant, the O.E. Wreoc-. The initial T^is repre-
sented in the usual Norman way with a parasitic vowel
between it and the r in Werecorddna, the spelling of this
name in a charter of William the Conqueror printed in the
Monasticon from an Inspeximus of Henry VI. In com-
pound names the Norman scribes usually represent wur by
or, so that Wreoc-wur^ine (dat. sing.) would be represented
by them as Werecordina. The name is written Worocordina
in a charter of Earl Hugh of Shrewsbury, 1094-1098,
printed in the Monasticon, iii, 520b, which represents the

6o All around the Wrekin.

Wroc- form. The original O.E. form must have been
TFreoce«-weor"5ign, which became by the eleventh century
Wreoce- by the weakening and dropping of the n in the
weak-accented syllable, and the Normans seem to have
failed to hear the resultant -e before \he wu ov weo, which
is not unnatural in such a polysyllabic word. But we have
traces of the persistance of this -e in late twelfth century
forms in the Pipe EoUs, which sometimes write the name
without it (probably as the result of dictation) and some-
times with it. The name is written Wrohe^lmr(Un in the
Roll for 21 Henry II, and in the chancellor's counterpart
for the 23 and 24 years. It is written with the Jc expressed
by the Norman ch as Wrochewv/r^n in the 18, 19 and 20
years. The syllable in question is entirely ignored in the
forms Wroch-wur^in, Wroc-wwr^in in the 22, 23 and 24
years, and in the first of Richard I.

Wroxeter similarly seems clearly to represent an O.E.
Wreocen-ceaster, reduced to Wreoce-ceaster. It is written
Bochecestre in Domesday, where ch has the usual Norman
value of k. The initial Wis represented in Wrochecestre
which occurs in an early twelfth century charter recited in
a confirmation of Henry III in the Monasiicon, iii, 522b,
and in the Wroccecestre of the Hundred Roll of 1255 cited
by Eyton. Through French influence cestre became pro-
nounced sesire, and so TFrocZ;eses<re easily becomes Wroxeter.

Wroxall, in the Isle of Wight, occurs in a Winchester
charter of 1038-1044 as Wrocces-heale (dat. sing.) in Kem-
ble. Codex Bvplomaticus, iv, 76. This Wrocc seems to be the
gen. of a masc. personal name. It also occurs in Wraxhall,
Wilts, Weroches-hale in Domesday ; Wroxton, co. Oxford,
in Domesday, Werochestane ; and Wraxall, Somerset, in
Domesday, Werocosale. Wroxham, Norfolk, and Wroxhall,
CO. Warwick, and Wroxhill, co. Bedford, seem to have
the same origin.

The name of Wrexham appears to be unconnected. It
occurs in a charter of 1236 as Wredesham (Calendar of
Charter Bolls, ii, p. 459;, and in 13 16 as Wryghtlesham
(Calendar of Close Bolls, p. 347^,

All around the Wrekin. 6i

APPENDIX II.

Edward Llwyd's Letter to the Rev. De. Mill,
Principal op Edmund Hall, Oxford.

Copied from the Oymmrodorion Record Series, No. 4, p. 410.
(See page 39 above.)

"Swansey, Sept. 14, [16] 96.
" Rev'd. Sir. I have here presum'd to trouble you with
a copy of an inscription,' which amongst several others I
met with this summer in North Wales. The monument
whence I took it was a stately pillar of very hard stone; of
the same kind with our common millstones. 'Twas of a
cylinder form; above twelve foot in height, seaven in cir-
cumference at the basis where it was thickest, and about
six near the top where smallest. The pedestal is a large
stone, five foot square and 15 inches thick; in the midst
whereof there's a round hole 12 inches deep wherein the
monument was placed. Within a foot of the top 'tis
encompassd with a round band or girth, resembling a cord ;
from whence 'tis square to the top, and each square adornd
with a ring, reaching from this band to the top and meet-
ing at the corners. It was erected on a small mount which
seems to have been cast up for that purpose ; but in the
late civil warres (or sooner) 'twas thrown down and broken
in several pieces, whence the inscription is so imperfect.
The reason I trouble you with it, is because I remember
amongst Usher's Letters one from Dr. Langbain to him,
wherein he writes to this purpose — ' I have received both the
inscriptions J and shall send you my thoughts of that a,t Vale
Grucis; but for the other, I give it over for desperat.' Now
this I send you is the IS. at Vale Crucis; and I doubt not,
but the vale receiv'd its name from this very stone, tho'
'twas never intended for a crosse. The copy Dr. Lang-
baine receiv'd was perhaps taken before the stone was
broke, and you may possibly meet with it amongst his

' This letter was printed also in the Cambro- Briton, 1820, pp. 55, 56,
where the editor appended the following footnote : — 'This inscription,
which from its imperfect state, it would be of very little use to tran-
scrihe here, Mr. Llwyd entitles "An Inscription at Maes y Groes, in
the parish of Llandysilio, in Denbighshire, transcribed anno 1696.'"

62 All around the Wrekin.

papers and letters, if you know where they are lodg'd ; or
direct me to search for it when I come to Oxford which
will be a month hence at farthest.

"The inscription would be legible enough were it entire.
It begins Concenn filius GaUeK, Gattel films BrochmaU,
Brochmal filius EUseg, EUseg filius Guoillauc. Concenn
itaque pronepos EUseg edificavit hunc la^idem proa/vo suo
EUseg &c- 'Tis remarkable that adjoyning to this monu-
ment there's a township calld Eglwysig, which name is
corrupted doubtlesse from this EUseg, th6 our greatest
critics interpret it Terra ecclesiasiica. Thus, in Caermardhin-
shire we find this epitaph : Servatour [pro servator] fidsei
patrieqwe semper amator Hie Paulinus jacit cultor pientis-
simus Bequi. The place where the stone lies is calld Pant
y Potion i.e., the Vale of Stakes, corruptly for Pant Powlin
Planities Paulini, I find other places denominated from
persons buryed at or near them; whence I gather they
were anciently men of great note, who had inscriptions on
their tombs be they never so rude and homely. But I
trouble you too much with trifles, so shall adde no more
but that I am,

"Worthy S"^, Your most obliged and humble servant,

"Edw. Lhwtd."

PosTSCEiPT : see p. 7.

My address in the Transactions of the Oxford Congress
for the History of Religions touches ground covered by this
paper : see II, 211, where I have suggested correcting
Eueyd into Eved, and equating it with Irish Ogma, Gaulish
Ogmios. The form required is Euvyd, which would be
written Ewwid or Euuyd : it occurs as Euuyd, and, misread,
as Mmyd. See Skene, ii, 200, 303, and Stephens' Gododin,
p. 377 {Ihmydd) ; also Skene, ii, 108, where it is leSyd, with
an intrusive i. The points of the equation are : (1) Gaulish
and Brythonic Ogmios was pronounced Ogmiios, and ii
makes yd in Welsh ; (2) gm or ghm behaves like Im, which
becomes Iv in Welsh, but remains Im in Irish; (3) Og or
ogh, becomes in Welsh ou, later eu. So Ogmios has its
exact equivalent in Euuyd in Welsh. Space fails me to
give analogies, to discuss texts, or draw conclusions.